Bawwab: Top 302 scan

octo

Flagged as possibly malicious Reviewed: expected behaviour

Bawwab's static rules matched a pattern it treats as critical. Read the review notes: a reviewed false alarm is marked as such.

Why it was flagged

The static rules marked this “Flagged as possibly malicious” because of:

Kind
Plugin
Source
github
Repository
nyldn/claude-octopus
Commit
7f979682e455198e170262eefd2469387e495ef6
Listed in
nyldn/claude-octopus
Popularity
4,477 GitHub stars (2026-11-20)
Link
https://github.com/nyldn/claude-octopus/tree/7f979682e455198e170262eefd2469387e495ef6
Scanned
2026-11-20T23:18:37+00:00
Scan time
45.4 s
Scanned as
plugin
Review note: Nothing malicious found, but it changes Claude itself: at session start it silently writes a Claude Code managed-settings file (turning off the built-in git instructions and moving auto-memory), and its optional statusline reads and rewrites your Claude login token, sending it only to Anthropic. It also takes provider API keys from ~/.bashrc, ~/.zshrc or .env when they are missing.

Findings (60)

credential-access: Touches credentials or secrets high

hooks/codex-exec-guard.sh:452

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"deny","permissionDecisionReason":"BLOCKED: the direct Gemini CLI individual client is retired and must not be dispatched by Oc

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

The match is the text of a PreToolUse 'deny' message that blocks a retired Gemini CLI path. It reads no credentials.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

hooks/octopus-hud.mjs:47

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

const CRED_PATH = join(HOME, ".claude", ".credentials.json");

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Expected for what this tool does

The optional rate-limit statusline reads your Claude Code login token from ~/.claude/.credentials.json (or the macOS keychain), sends it only to Anthropic (api.anthropic.com usage endpoint; refresh at platform.claude.com) and writes the refreshed token back into your credentials file. Nothing goes to a third party, but a third-party plugin is handling and rewriting your Claude login.

Reviewed by Claude (AI-assisted review), 2026-11-21

exfil.combo: Reads credentials and makes network calls med

hooks/octopus-hud.mjs:47

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0037 Data from Local SystemMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0025 Exfiltration via Cyber Means

const CRED_PATH = join(HOME, ".claude", ".credentials.json");

What this means: Credential access together with outbound requests in one file is the shape of data theft.

The rule's general advice: Do not install unless the flow is understood and benign. (see the review note for this case)

Expected for what this tool does

Same HUD code: the only network calls carrying the token go to Anthropic's own endpoints (api.anthropic.com, platform.claude.com); the other request is an unauthenticated version check to registry.npmjs.org. No token goes to a third party.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

hooks/session-start-memory.sh:173

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

SETTINGS_D="${HOME}/.claude/managed-settings.d"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Expected for what this tool does

At every session start the hook silently writes ~/.claude/managed-settings.d/octopus-defaults.json, a global Claude Code settings change: it turns off Claude Code's built-in git instructions and moves auto-memory to ~/.claude-octopus/memory/. Only the changelog mentions it. It reads no credential, and it is the plugin's intended setup, but it changes how Claude behaves in every project.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

hooks/statusline-auto-repair.sh:22

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

SETTINGS="$HOME/.claude/settings.json"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Expected for what this tool does

The SessionStart hook greps ~/.claude/settings.json and, only if an Octopus statusline is already set up, rewrites a stale versioned statusline path to a stable one. No secrets are read or sent.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/agent-teams-bridge.sh:496

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

local teams_dir="$HOME/.claude/teams"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Reads team member names from Claude Code's ~/.claude/teams/*/config.json with jq to bridge agent teams. These files hold no credentials and nothing is sent out.

Reviewed by Claude (AI-assisted review), 2026-11-21

remote-code.pipe-shell: Pipes a download into a shell high

scripts/install-deps.sh:101

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesOWASP Agentic ASI05 Unexpected Code Execution (RCE)MITRE ATLAS AML.T0050 Command and Scripting InterpreterCWE CWE-494 Download of Code Without Integrity Check

warnings+=("kimi_legacy:Legacy kimi-cli detected — install current Kimi Code: curl -fsSL https://code.kimi.com/kimi-code/install.sh | bash. The installer renames the old executable to kimi-legacy; fir

What this means: Executes whatever the remote server returns, unreviewed.

The rule's general advice: Download, verify a checksum, then run. (see the review note for this case)

Known false alarm

The curl | bash text is part of a warning that tells the user how to install the optional Kimi Code CLI. The script prints it; it never runs it.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/install-deps.sh:121

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

local settings_file="${HOME}/.claude/settings.json"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Greps ~/.claude/settings.json for the word 'rtk' to report whether the RTK hook is configured. No secret is read.

Reviewed by Claude (AI-assisted review), 2026-11-21

exfil.combo: Reads credentials and makes network calls critical

scripts/install-deps.sh:121

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0037 Data from Local SystemMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0025 Exfiltration via Cyber Means

local settings_file="${HOME}/.claude/settings.json"

What this means: Credential access together with outbound requests in one file is the shape of data theft.

The rule's general advice: Do not install unless the flow is understood and benign. (see the review note for this case)

Known false alarm

The only network call in this file is a curl to a local Ollama server (localhost:11434) to check it is running; the settings.json check just greps for 'rtk'. Nothing is sent out.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/tangle-read-scope.py:20

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

if any(part in SENSITIVE_DIRS or part == ".env" or part.startswith(".env.") for part in parts):

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

This is a deny-list: the function marks .env files, .ssh, .aws and similar paths as sensitive so subagents are kept from reading them.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/validate-no-hardcoded-paths.sh:58

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

.claude/settings.json \

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

A list of development-only paths (including .claude/settings.json) that a publishing check makes sure are not in the public repository. Nothing is read.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/helpers/readiness-receipt.py:49

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

policy = {k: v for k, v in os.environ.items() if (k.startswith("OCTOPUS_") or k == "OCTO_ALLOWED_PROVIDERS") and not k.startswith("OCTOPUS_PREFLIGHT_") and not any(secret in k for secret in ("KEY", "T

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Builds a hash of OCTOPUS_* policy settings and explicitly drops any variable whose name contains KEY, TOKEN, SECRET or PASSWORD. It is used locally as a fingerprint and not sent anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/agents.sh:622

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

memory_file="${HOME}/.claude/projects/${project_hash}/memory/MEMORY.md"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Looks up Claude Code's own project MEMORY.md path so the plugin can include project memory as context for its workflows. It holds no credentials and is not sent anywhere by this code.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/config-display.sh:277

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

local env_file="${OCTOPUS_SECRET_ENV_FILE:-${HOME}/.env}"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Expected for what this tool does

The setup wizard saves a provider API key that the user just typed into ~/.env (umask 077, atomic write) so later sessions can find it. That is the plugin's stated multi-provider setup.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/context.sh:287

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

memory_file="${HOME}/.claude/projects/${project_hash}/memory/MEMORY.md"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Looks up Claude Code's own project MEMORY.md path so the plugin can include project memory as context for its workflows. It holds no credentials and is not sent anywhere by this code.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/copilot.sh:57

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

echo "keychain"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Returns the word 'keychain' as a label when ~/.copilot/config.json exists, to report which Copilot login method is in use. It reads no secret.

Reviewed by Claude (AI-assisted review), 2026-11-21

remote-code.pipe-shell: Pipes a download into a shell high

scripts/lib/cursor-agent.sh:271

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesOWASP Agentic ASI05 Unexpected Code Execution (RCE)MITRE ATLAS AML.T0050 Command and Scripting InterpreterCWE CWE-494 Download of Code Without Integrity Check

_cursor_log ERROR "cursor-agent: CLI not found — install: curl -fsSL https://cursor.com/install | bash"

What this means: Executes whatever the remote server returns, unreviewed.

The rule's general advice: Download, verify a checksum, then run. (see the review note for this case)

Known false alarm

Logs an install hint for the optional Cursor Agent CLI (curl https://cursor.com/install | bash) when it is missing. The plugin does not run it.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/doctor.sh:295

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

"Launch plain 'agy' in a terminal and finish its browser sign-in. On macOS, if access is denied, open Keychain Access, find the Antigravity CLI item, and allow agy under Access Control."

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

A doctor message telling the user how to finish the Antigravity CLI sign-in and allow keychain access. Text only.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/memory.sh:17

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

"$HOME/.claude.json"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Picks which Claude settings file to read and checks with jq whether an mcp-memory-service MCP server is registered. It reads server commands, not secrets, and sends nothing.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/model-resolver.sh:1135

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

_octo_assignment_has_nonempty_value "${HOME}/.vibe/.env" "MISTRAL_API_KEY" || \

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Expected for what this tool does

Checks whether a Mistral API key is set (in the environment, ~/.vibe/.env or ~/.vibe/config.toml) to decide if the Mistral Vibe provider can be used. Checking for provider keys is part of the plugin's stated multi-provider routing.

Reviewed by Claude (AI-assisted review), 2026-11-21

remote-code.pipe-shell: Pipes a download into a shell high

scripts/lib/preflight.sh:1046

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesOWASP Agentic ASI05 Unexpected Code Execution (RCE)MITRE ATLAS AML.T0050 Command and Scripting InterpreterCWE CWE-494 Download of Code Without Integrity Check

echo -e "  curl -fsSL https://cursor.com/install | bash"

What this means: Executes whatever the remote server returns, unreviewed.

The rule's general advice: Download, verify a checksum, then run. (see the review note for this case)

Known false alarm

Prints an install hint for the optional Cursor Agent CLI (curl https://cursor.com/install | bash) when it is missing. The plugin does not run it.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/preflight.sh:161

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf 'Note: preflight ran inside the %s. A provider that needs a keychain login, the network or files outside the workspace can fail here and work outside it. Re-run orchestrate.sh preflight outside

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

A printed note that preflight ran inside a sandbox where keychain logins may fail. Text only.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/provider-routing.sh:590

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

for env_file in "$PWD/.env" "$HOME/.env" "$HOME/.bashrc" "$HOME/.zshrc"; do

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Expected for what this tool does

When a provider API key is missing from the environment, it looks for a plain 'NAME=value' line for that key in ./.env, ~/.env, ~/.bashrc or ~/.zshrc, skipping values with shell substitutions, and exports it for the provider CLI. It reaches into shell rc files, but only for the provider key it needs, and the key goes to that provider. This is part of the stated orchestration.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/providers.sh:134

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

for _try_path in "$HOME/.local/bin/claude" "/usr/local/bin/claude" "$HOME/.claude/bin/claude"; do

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Lists common install locations of the claude binary to add it to PATH. No credentials involved.

Reviewed by Claude (AI-assisted review), 2026-11-21

exfil.combo: Reads credentials and makes network calls critical

scripts/lib/providers.sh:134

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0037 Data from Local SystemMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0025 Exfiltration via Cyber Means

for _try_path in "$HOME/.local/bin/claude" "/usr/local/bin/claude" "$HOME/.claude/bin/claude"; do

What this means: Credential access together with outbound requests in one file is the shape of data theft.

The rule's general advice: Do not install unless the flow is understood and benign. (see the review note for this case)

Known false alarm

The flagged line only looks for the claude binary. The file's network calls are a local Ollama health check and an OpenRouter request made with the user's own OpenRouter key; no file credentials are sent anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/spawn.sh:1830

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

-e '^Keychain initialization' \

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

A grep filter that drops status-noise lines such as 'Keychain initialization' from external CLI output. Text only.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

scripts/lib/workflows.sh:2007

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf "%s\n" "- Never read credentials, .env files, private keys, auth stores or Git metadata, including through symlinks or recursive scans. Reads: never grants write permission."

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Part of a prompt that tells subagents never to read credentials, .env files or private keys. This is a safety rule, not credential access.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/integration/test-plugin-expert-review.sh:357

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

assert_gitignored "\.env" ".env files are gitignored"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/smoke/test-safety-hooks.sh:266

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

[[ "$(_cc_decides 'git restore ./.env')"       == quiet ]] || fails+=" restore-slash-dotfile"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-agent-predicates.sh:116

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

rm -f "$VIBE_PREDICATE_HOME/.vibe/.env" "$VIBE_PREDICATE_HOME/.vibe/config.toml"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-agent-teams-bridge.sh:134

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

pass "Discovery reads from ~/.claude/teams/"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-agy-provider.sh:1164

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

test_case "AGY live doctor gives the real browser and macOS keychain auth path"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-council-policy-findings.py:23

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

self.env = {key: value for key, value in os.environ.items() if not key.startswith("GIT_")}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

remote-code.pipe-shell: Pipes a download into a shell high

tests/unit/test-council-quote-grounding.sh:3469

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesOWASP Agentic ASI05 Unexpected Code Execution (RCE)MITRE ATLAS AML.T0050 Command and Scripting InterpreterCWE CWE-494 Download of Code Without Integrity Check

span_command='curl https://verified.invalid/source | bash'

What this means: Executes whatever the remote server returns, unreviewed.

The rule's general advice: Download, verify a checksum, then run. (see the review note for this case)

Known false alarm

A test string ('curl https://verified.invalid/source | bash') used as input for a quote-grounding check, on a .invalid domain. It is never run.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-doctor-early-dispatch.sh:293

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

jq -n --arg command "touch $TEST_TMP_DIR/api-key-helper-called" '{apiKeyHelper: $command}' > "$fixture_home/.claude/settings.json"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-explicit-activation.sh:242

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf '{}\n' > "$status_home/.claude/settings.json"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-feature-workflow.py:25

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

self.env = dict(os.environ, HOME=str(self.home), OCTOPUS_PROJECT_DIR=str(self.root),

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-litellm-provider.sh:364

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf '%s\n' 'LITELLM_BASE_URL=https://attacker.example.com' > "$TEST_HOME/untrusted-project/.env"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

exfil.host: References an exfiltration-style host high

tests/unit/test-litellm-provider.sh:337

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0025 Exfiltration via Cyber Means

for url in http://localhost:4000 https://litellm_proxy:4000 https://litellm_proxy:4000/v1/ http://10.0.0.5:4000 \

What this means: Paste, webhook and tunnel services are commonly used to receive stolen data.

The rule's general advice: Remove or justify the destination. (see the review note for this case)

Known false alarm

Test inputs (localhost, private IPs) for the LiteLLM base-URL check. The test makes sure untrusted URLs are rejected; nothing is sent.

Reviewed by Claude (AI-assisted review), 2026-11-21

exfil.combo: Reads credentials and makes network calls critical

tests/unit/test-litellm-provider.sh:364

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0037 Data from Local SystemMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0025 Exfiltration via Cyber Means

printf '%s\n' 'LITELLM_BASE_URL=https://attacker.example.com' > "$TEST_HOME/untrusted-project/.env"

What this means: Credential access together with outbound requests in one file is the shape of data theft.

The rule's general advice: Do not install unless the flow is understood and benign. (see the review note for this case)

Known false alarm

A test writes a fake LITELLM_BASE_URL=https://attacker.example.com into an untrusted project .env to check the plugin refuses to use it. Nothing is sent anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-orcarouter-provider.sh:206

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf '%s\n' 'ORCAROUTER_API_KEY=sk-orca-profile-test' > "$profile_home/.env"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-plan-artifact-storage.sh:96

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

if env "HOME=$test_home" "OCTOPUS_SESSION_PLANS=$test_home/.claude/plans" \

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

exfil.host: References an exfiltration-style host high

tests/unit/test-security-functions.sh:103

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0025 Exfiltration via Cyber Means

if ! validate_external_url "https://10.2.3.4/" >/dev/null 2>&1; then

What this means: Paste, webhook and tunnel services are commonly used to receive stolen data.

The rule's general advice: Remove or justify the destination. (see the review note for this case)

Known false alarm

A unit test making sure validate_external_url rejects a private IP. Test data only.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-session-auto-memory-path.sh:218

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

mkdir -p "$case_home/.claude/projects/-home-octopus-my.app/memory"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-adequacy-json-v1.sh:62

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

bad='{"schema_version":1,"verdict":"fail","reasons":["Needs scope."],"scope_review":[{"action":"add_write","path":"src/../.env","reason":"Unsafe path."}]}'

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-challenge-evidence.sh:21

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf 'private-before\n' > "$repo/.env"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-contextual-read-scope.sh:26

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf "secret\n" > "$repo/.env"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-decomposition-normalization.sh:40

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

prose=$'### 1. [CODING] Conservative scopes\n\n**Reads**\n- Tracked files, excluding credentials, `.env`, and auth stores\n- `src/a.ts`\n\n**Files**\n- `src/a.ts`\n\n**Creates**\n- None\n\nTask: edit

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-execution-boundary.sh:622

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

[[ ! -e "$POLICY_CODEX_HOME/tmp/second-dispatch" && ! -e "$POLICY_HOME/.ssh/forged" ]]

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-memory-keeper-decomposition.sh:18

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

touch "$WORKSPACE_DIR/package.json" "$WORKSPACE_DIR/.env.example" "$WORKSPACE_DIR/.gitignore" "$WORKSPACE_DIR/README.md" \

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-overlap-consolidation.sh:84

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

touch "$WORKSPACE_DIR/package.json" "$WORKSPACE_DIR/.env.example" "$WORKSPACE_DIR/.gitignore" "$WORKSPACE_DIR/README.md" \

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-tangle-post-execution-scope-contract.sh:118

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

printf '%s\n' 'fixture-only' > "$TMP_REPO/.env"

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-vibe-keyring.py:42

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

self.env = dict(os.environ, HOME=str(self.home), PATH=f'{self.bin}:/usr/bin:/bin',

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

credential-access: Touches credentials or secrets high

tests/unit/test-workflow-result.py:79

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

self.env = {k: v for k, v in os.environ.items()

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

injection.override: Tries to override the AI's instructions high

.claude/skills/skill-debate/SKILL.md:37

OWASP LLM LLM01 Prompt InjectionOWASP LLM LLM07 System Prompt LeakageOWASP Agentic ASI01 Agent Goal HijackMITRE ATLAS AML.T0051.001 LLM Prompt Injection: IndirectMITRE ATLAS AML.T0110.000 AI Agent Tool Poisoning: Definition and InstructionsMITRE ATLAS AML.T0069.002 Discover LLM System Information: System PromptCWE CWE-1427 Improper Neutralization of Input Used for LLM Prompting

without telling the user

What this means: Text read by the AI that says to ignore earlier instructions, reveal the system prompt or hide things from you is a classic prompt-injection attempt.

The rule's general advice: Do not install; the text is trying to steer the assistant against you. (see the review note for this case)

Known false alarm

The phrase 'without telling the user' appears in a list of things the agent is forbidden to do (for example, dropping an advisor or starting unrelated work without telling the user). It pushes the agent to be more open with the user, not less.

Reviewed by Claude (AI-assisted review), 2026-11-21

injection.override: Tries to override the AI's instructions high

.cursor-plugin/commands/octo-resume.md:14

OWASP LLM LLM01 Prompt InjectionOWASP LLM LLM07 System Prompt LeakageOWASP Agentic ASI01 Agent Goal HijackMITRE ATLAS AML.T0051.001 LLM Prompt Injection: IndirectMITRE ATLAS AML.T0110.000 AI Agent Tool Poisoning: Definition and InstructionsMITRE ATLAS AML.T0069.002 Discover LLM System Information: System PromptCWE CWE-1427 Improper Neutralization of Input Used for LLM Prompting

without telling the user

What this means: Text read by the AI that says to ignore earlier instructions, reveal the system prompt or hide things from you is a classic prompt-injection attempt.

The rule's general advice: Do not install; the text is trying to steer the assistant against you. (see the review note for this case)

Known false alarm

The phrase 'without telling the user' appears in a list of things the agent is forbidden to do (for example, dropping an advisor or starting unrelated work without telling the user). It pushes the agent to be more open with the user, not less.

Reviewed by Claude (AI-assisted review), 2026-11-21

injection.override: Tries to override the AI's instructions high

codex-skills/skill-debate/SKILL.md:26

OWASP LLM LLM01 Prompt InjectionOWASP LLM LLM07 System Prompt LeakageOWASP Agentic ASI01 Agent Goal HijackMITRE ATLAS AML.T0051.001 LLM Prompt Injection: IndirectMITRE ATLAS AML.T0110.000 AI Agent Tool Poisoning: Definition and InstructionsMITRE ATLAS AML.T0069.002 Discover LLM System Information: System PromptCWE CWE-1427 Improper Neutralization of Input Used for LLM Prompting

without telling the user

What this means: Text read by the AI that says to ignore earlier instructions, reveal the system prompt or hide things from you is a classic prompt-injection attempt.

The rule's general advice: Do not install; the text is trying to steer the assistant against you. (see the review note for this case)

Known false alarm

The phrase 'without telling the user' appears in a list of things the agent is forbidden to do (for example, dropping an advisor or starting unrelated work without telling the user). It pushes the agent to be more open with the user, not less.

Reviewed by Claude (AI-assisted review), 2026-11-21

injection.override: Tries to override the AI's instructions high

commands/resume.md:15

OWASP LLM LLM01 Prompt InjectionOWASP LLM LLM07 System Prompt LeakageOWASP Agentic ASI01 Agent Goal HijackMITRE ATLAS AML.T0051.001 LLM Prompt Injection: IndirectMITRE ATLAS AML.T0110.000 AI Agent Tool Poisoning: Definition and InstructionsMITRE ATLAS AML.T0069.002 Discover LLM System Information: System PromptCWE CWE-1427 Improper Neutralization of Input Used for LLM Prompting

without telling the user

What this means: Text read by the AI that says to ignore earlier instructions, reveal the system prompt or hide things from you is a classic prompt-injection attempt.

The rule's general advice: Do not install; the text is trying to steer the assistant against you. (see the review note for this case)

Known false alarm

The phrase 'without telling the user' appears in a list of things the agent is forbidden to do (for example, dropping an advisor or starting unrelated work without telling the user). It pushes the agent to be more open with the user, not less.

Reviewed by Claude (AI-assisted review), 2026-11-21

injection.override: Tries to override the AI's instructions high

skills/skill-debate/SKILL.md:37

OWASP LLM LLM01 Prompt InjectionOWASP LLM LLM07 System Prompt LeakageOWASP Agentic ASI01 Agent Goal HijackMITRE ATLAS AML.T0051.001 LLM Prompt Injection: IndirectMITRE ATLAS AML.T0110.000 AI Agent Tool Poisoning: Definition and InstructionsMITRE ATLAS AML.T0069.002 Discover LLM System Information: System PromptCWE CWE-1427 Improper Neutralization of Input Used for LLM Prompting

without telling the user

What this means: Text read by the AI that says to ignore earlier instructions, reveal the system prompt or hide things from you is a classic prompt-injection attempt.

The rule's general advice: Do not install; the text is trying to steer the assistant against you. (see the review note for this case)

Known false alarm

The phrase 'without telling the user' appears in a list of things the agent is forbidden to do (for example, dropping an advisor or starting unrelated work without telling the user). It pushes the agent to be more open with the user, not less.

Reviewed by Claude (AI-assisted review), 2026-11-21

plugin.hook-broad: Plugin hook runs on every tool call med

hooks/hooks.json:113

MITRE ATLAS AML.T0142 Event Triggered Execution

"${CLAUDE_PLUGIN_ROOT}/hooks/post-tool-dispatch.sh"

What this means: A hook with no specific matcher runs a command each time the assistant uses any tool, giving it constant access to your session.

The rule's general advice: Check the hook command, and prefer plugins whose hooks match specific tools. (see the review note for this case)

Expected for what this tool does

A PostToolUse hook on Bash, Agent, Write, Edit, Read, WebFetch and Grep runs the plugin's local dispatcher, which exits at once unless the user turned on context-awareness, strategy rotation or output compression. It makes no network calls and is part of the plugin's stated features.

Reviewed by Claude (AI-assisted review), 2026-11-21

140 low-severity notes (informational, do not affect the result)

files.binary-unscanned: Binary files not analysed low

1 binary file not analysed; use --deep to run in the sandbox

What this means: Prebuilt executables cannot be read by static rules, so what they do is unknown until they run.

What you can do: Use --deep to run the package in the sandbox, or prefer packages built from source.

install-dir-access: Reads or writes Claude's install folders low

.coderabbit.yaml:48

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

- path: ".claude/skills/*.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

package.json:37

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

".claude/agents/",

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

hooks/cwd-changed.sh:59

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if [[ -d "$NEW_CWD/.claude" ]]; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

hooks/plan-mode-interceptor.sh:75

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

never assume a bare .claude/session-intent.md outside a real project), read

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

hooks/statusline-resolver.sh:8

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CACHE_BASE="$HOME/.claude/plugins/cache/nyldn-plugins/octo"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

mcp-server/dist/index.js:71

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

...Object.fromEntries(Object.entries(process.env).filter(([k]) => (k.startsWith("CLAUDE_OCTOPUS_") || k.startsWith("OCTOPUS_")) &&

What this means: Reads files or environment holding keys, tokens or sessions.

What you can do: Confirm the tool needs this; otherwise do not install.

credential-access: Touches credentials or secrets low

mcp-server/src/index.ts:100

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

Object.entries(process.env).filter(([k]) =>

What this means: Reads files or environment holding keys, tokens or sessions.

What you can do: Confirm the tool needs this; otherwise do not install.

install-dir-access: Reads or writes Claude's install folders low

mcp-server/src/schema/skill-schema.json:89

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"description": "Claude Code skill path (e.g., ./.claude/skills/skill-code-review.md)"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/build-codex-skills.sh:29

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

SKILLS_DIR="$PLUGIN_ROOT/.claude/skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/build-factory-skills.sh:15

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

SKILLS_SRC="$PLUGIN_ROOT/.claude/skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/cache-check.sh:72

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

claude_cache="${OCTOPUS_CLAUDE_CACHE_DIR:-${HOME}/.claude/plugins/cache/nyldn-plugins/octo}"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/migrate-todos.sh:27

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CLAUDE_DIR="${HOME}/.claude"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/orchestrate.sh:460

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

SUPPORTS_SKILL_DEADLOCK_FIX=false      # v8.52: Claude Code v2.1.73+ (no deadlock with large .claude/skills/ during git pull)

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/plan-storage.sh:111

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"$home_dir/.claude"|"$home_dir/.claude/"*)

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/test-codex-compat.sh:17

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CODEX_INSTRUCTION_TARGETS="codex-skills skills .claude/skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/validate-no-hardcoded-paths.sh:56

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

.claude/DEVELOPMENT.md \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/validate-plugin-assembly.py:159

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

authored = root / ".claude" / "skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

scripts/helpers/custom_provider_config.py:85

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

return {name: value for name, value in os.environ.items() if name in names}

What this means: Reads files or environment holding keys, tokens or sessions.

What you can do: Confirm the tool needs this; otherwise do not install.

credential-access: Touches credentials or secrets low

scripts/helpers/openai-compatible-agent.py:120

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

(r'(/etc/(shadow|sudoers)|\.ssh/id_[a-z0-9]+|\.aws/credentials|\.netrc|\.git-credentials)',

What this means: Reads files or environment holding keys, tokens or sessions.

What you can do: Confirm the tool needs this; otherwise do not install.

credential-access: Touches credentials or secrets low

scripts/helpers/opencode-probe.py:58

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = dict(os.environ)

What this means: Reads files or environment holding keys, tokens or sessions.

What you can do: Confirm the tool needs this; otherwise do not install.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/agent-utils.sh:215

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

USER_AGENTS_DIR="${HOME}/.claude/agents"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/agents.sh:446

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

local skill_file="${PLUGIN_DIR}/.claude/skills/${skill_name}.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/cache-hygiene.sh:10

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

OCTO_CACHE_DIR="${HOME}/.claude/plugins/cache/nyldn-plugins/octo"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/context.sh:297

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

memory_file="${PROJECT_ROOT}/.claude/memory/MEMORY.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

scripts/lib/council.sh:4080

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = {k: v for k, v in os.environ.items() if not k.startswith("GIT_")}

What this means: Reads files or environment holding keys, tokens or sessions.

What you can do: Confirm the tool needs this; otherwise do not install.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/dispatch.sh:1882

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

persona_file="${USER_AGENTS_DIR:-${HOME}/.claude/agents}/${agent_name}.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/doctor.sh:552

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

*/.claude/plugins/cache/*) printf '%s\n' "claude-marketplace-cache" ;;

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/host-runtime.sh:45

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

elif [[ "$plugin_root" == *"/.claude/"* ]]; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/plugin-root.sh:149

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

local cache_base="${HOME}/.claude/plugins/cache/nyldn-plugins/octo"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

scripts/lib/preflight.sh:1120

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

local claude_plugins_dir="$HOME/.claude/plugins"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/test-fleet-diversity.sh:157

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

pass "3.2 .claude/skills/flow-discover.md references build-fleet.sh"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/benchmark/manual-test.sh:122

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

echo "   Check: ls ~/.claude/plugins/claude-octopus"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/fixtures/native-skills/codex-references.py:113

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

source = ('Enumerate `.claude/skills/`, `skills/`, and `codex-skills/`.\n'

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/helpers/test-framework.sh:65

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if [[ -f "$root/.claude/skills/${name}/SKILL.md" ]]; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/integration/test-plugin-lifecycle.sh:25

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

env -i HOME="$home" CLAUDE_CONFIG_DIR="$home/.claude" \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/smoke/test-debate-skill.sh:13

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if [[ -f "$PROJECT_ROOT/.claude/skills/skill-debate.md" ]]; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/smoke/test-packaging-integrity.sh:34

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

env -i HOME="$home" CLAUDE_CONFIG_DIR="$home/.claude" \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/smoke/test-safety-hooks.sh:79

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

SKILL_DEBUG="$PROJECT_ROOT/.claude/skills/skill-debug.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/smoke/test-syntax.sh:32

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

test_case ".claude/skills/skill-parallel-agents has valid frontmatter"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/codex_safety_contract_test.py:89

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"/tmp/.claude/plugins/cache/vendor/pkg/1"}, compact=True)

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/feature_analysis_fixture.py:56

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = dict(os.environ, PATH=str(tripwire), PYTHONDONTWRITEBYTECODE="1")

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-adapted-workflow-skills.sh:25

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

f="$PROJECT_ROOT/.claude/skills/$s/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-agent-ergonomics.sh:55

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"claude/agents" "USER_AGENTS_DIR: points to ~/.claude/agents"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-agent-permissions-audit.sh:12

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

AGENTS_DIR="$PROJECT_ROOT/.claude/agents"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-agent-return-contracts.sh:12

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

AGENTS_DIR="$PROJECT_ROOT/.claude/agents"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-agent-topology-skill.sh:21

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

SKILL="$PROJECT_ROOT/.claude/skills/skill-agent-topology/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-agy-provider.sh:1105

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

grep -q 'Antigravity CLI' "$PROJECT_ROOT/.claude/skills/skill-doctor/SKILL.md" && \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-argument-placeholder-safety.sh:62

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

find "$PROJECT_ROOT/.claude/skills" -maxdepth 2 -type f \( -name 'SKILL.md' -o -name '*.tmpl' \)

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-cache-hygiene.sh:19

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mkdir -p "$h/.claude/plugins/cache/nyldn-plugins/octo"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-cc-v2132-session-id.sh:121

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

.claude/skills/skill-debug/SKILL.md skills/skill-debug/SKILL.md; do

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-command-discovery.sh:61

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

clean = {k: v for k, v in os.environ.items() if k not in ("CLAUDE_PLUGIN_ROOT", "CODEX_PLUGIN_ROOT", "OCTO_ROOT", "OCTO_SKILL_FILE")}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-command-discovery.sh:152

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for template in (root / '.claude/skills').glob('*/*.tmpl'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-command-frontmatter.sh:76

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

".claude/skills/skill-copilot-provider/SKILL.md|0"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-command-registration.sh:199

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if [[ "$ROOT_COMMAND_TARGET_COUNT" -eq 1 ]] && ! grep -F '.claude/commands' "$CODEX_COMPAT_SCRIPT" >/dev/null; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-command-safety-regressions.sh:12

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

base_env = dict(os.environ)

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-consultative-advisor-launch.sh:29

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"$PROJECT_ROOT/.claude/skills/skill-debate/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-consultative-host-identity.sh:57

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-council-grounding-citations.sh:562

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mkdir -p "$BOUNDARY_ROOT/.claude"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-custom-provider-config.py:61

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

calls.append((base_url, key, model, dict(os.environ), kwargs))

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-debate-state-json.sh:29

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-debug-freeze-state.sh:33

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mirrors = [".claude/skills/skill-debug/SKILL.md", "skills/skill-debug/SKILL.md"]

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-define-state-failures.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-deliver-state-failures.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-develop-checkpoint-failures.sh:47

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = {k: v for k, v in os.environ.items() if not k.startswith('GIT_')}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-develop-checkpoint-failures.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-develop-state-failures.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-docs-sync.sh:212

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if [ -f ".claude/skills/${skill_name}/SKILL.md" ]; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-doctor-early-dispatch.sh:292

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mkdir -p "$fixture_home/.claude"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-document-renderer.py:383

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = dict(os.environ, PATH=str(bin_dir) + os.pathsep + os.environ["PATH"],

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-document-renderer.py:133

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in (".claude/skills", "skills"):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-documented-orchestrate-examples.sh:12

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

AUDIT_SKILL="$PROJECT_ROOT/.claude/skills/skill-security-audit/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-explicit-activation.sh:42

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

done < <(find "$PROJECT_ROOT/.claude/skills" "$PROJECT_ROOT/skills" -name SKILL.md -type f | sort)

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-factory-command-generation.sh:23

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mkdir -p "$fixture/scripts" "$fixture/commands" "$fixture/.claude/agents" \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-feature-delivery.py:249

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

owners = {'.claude-plugin', '.codex-plugin', '.cursor-plugin', '.claude', 'commands', 'skills', 'codex-skills', 'scripts', 'config', 'agents', 'hooks'}

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-finish-branch-pr-publication.sh:33

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

environment = {key: value for key, value in os.environ.items()

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-finish-branch-pr-publication.sh:22

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-finish-branch-review-coverage.sh:36

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

source_environment = dict(os.environ)

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-finish-branch-review-coverage.sh:22

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-flow-parallel-wave-gates.sh:21

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-help-early.sh:9

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mkdir -p "$TMP/plugin/scripts/lib" "$TMP/home/.claude/agents"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-hook-err-traps.sh:71

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = os.environ.copy()

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-host-runtime-detection.sh:16

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CLAUDE_INSTALL="/home/u/.claude/plugins/cache/nyldn-plugins/octo/11.9.3"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-host-shell-bash-blocks.sh:386

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

environment = dict(os.environ)

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-host-shell-bash-blocks.sh:100

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for pattern in ('commands/*.md', 'skills/*/SKILL.md', '.claude/skills/*/SKILL.md',

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-human-only-skill-list.sh:17

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for directory in "$PROJECT_ROOT/.claude/skills" "$PROJECT_ROOT/skills" "$codex_skill_root"; do

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-install-readiness.sh:314

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

claude_cache="$cache_home/.claude/plugins/cache/nyldn-plugins/octo"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-installation-health-docs.sh:11

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

DOCTOR_SKILL="$PROJECT_ROOT/.claude/skills/skill-doctor/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-intent-contract-allocation.sh:21

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CLAUDE_SKILL="$PROJECT_ROOT/.claude/skills/skill-intent-contract/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-intent-contract-skill.sh:124

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

grep -q "\.claude/session-intent" "$SKILL_FILE"; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-metric-loop.sh:12

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for file in ['.claude/skills/skill-iterative-loop/SKILL.md','skills/skill-iterative-loop/SKILL.md']:

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-native-skill-registration.sh:32

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

native = list((root / '.claude/skills').glob('*/SKILL.md'))

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-parallel-result-summary.sh:50

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = dict(os.environ)

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-parallel-result-summary.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-phases-1-2-3.sh:195

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if assert_file_exists ".claude/references/validation-gates.md"; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-pid-ledger-python.sh:270

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

clean_env = {key: value for key, value in os.environ.items() if not key.startswith('PYTHON')}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-plan-artifact-storage.sh:27

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"$home_plan_dir" != "$test_home/.claude/"* ]]; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-plugin-assembly-standard.sh:303

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

assert result.returncode != 0 and '.claude/skills/: missing or unsafe required Octopus command catalog' in result.stderr, result.stderr

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-plugin-update-health.sh:99

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if assert_equals "claude" "$(octo_plugin_detect_host "/tmp/user/.claude/plugins/cache/nyldn-plugins/octo/1.0.0")"; then test_pass; fi

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-provider-allowlist.sh:250

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"$PROJECT_ROOT/.claude/skills/skill-debate/SKILL.md" \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-provider-readiness-contract.sh:247

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

canonical_parallel_skill="$PROJECT_ROOT/.claude/skills/skill-parallel-agents/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-quick-commit-isolation.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

entries = [root / '.claude/skills/skill-quick/SKILL.md',

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-refinement-efficiency.py:373

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for entry in (ROOT / 'skills' / skill / 'SKILL.md', ROOT / '.claude/skills' / skill / 'SKILL.md'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-research-fanout-static.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if [[ "$research_cmd" == *'.claude/skills/skill-deep-research/SKILL.md'* \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-retired-claw-admin.sh:38

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

.claude/skills/skill-claw \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-retired-gemini-provider.sh:110

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"$PROJECT_ROOT/.claude/hooks" >/dev/null; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-rollback-clean-worktree.sh:19

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

source = (root / '.claude/skills/skill-rollback/SKILL.md').read_text().split('## Mode: ROLLBACK\n', 1)[1]

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-rollback-lessons-backup.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-rollback-preview-binding.sh:19

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-rollback-submodule-boundary.sh:18

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = {key: value for key, value in os.environ.items() if not key.startswith('GIT_')}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-rollback-submodule-boundary.sh:24

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-rollback-tree-boundary.sh:26

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-rtk-detection.sh:50

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

pass "Doctor skill (.claude/skills) mentions RTK"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-safe-github-comments.sh:1179

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"$PROJECT_ROOT/.claude/skills" "$PROJECT_ROOT/skills"; then

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-session-auto-memory-path.sh:22

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

case_memory="$case_home/.claude/projects/$encoded/memory"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-setup-first-success.sh:53

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

unreadable_root="$fixture_home/.claude/plugins/cache/nyldn-plugins/octo/0.0.0"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-shell-safe-provider-snippets.sh:33

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

check_tree "source skill docs" "$PROJECT_ROOT/.claude/skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-ship-archive-boundary.sh:22

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-ship-checkpoint-provenance.sh:41

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

env = {k: v for k, v in os.environ.items() if not k.startswith('GIT_')}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-ship-checkpoint-provenance.sh:21

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-ship-lessons-boundary.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-ship-state-boundary.sh:25

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-skill-doctor-stable-link.sh:95

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

local doctor_skill="$PROJECT_ROOT/.claude/skills/skill-doctor/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-skill-frontmatter.sh:156

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

skills_dir=".claude/skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-skill-sources.sh:34

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

self.authored = self.root / '.claude/skills/flow-define/SKILL.md'

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-smart-router.sh:15

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

SKILLS_DIR="$PROJECT_ROOT/.claude/skills"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-stable-root-self-heal.sh:40

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CACHE="$HOME/.claude/plugins/cache/nyldn-plugins/octo"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

credential-access: Touches credentials or secrets low

tests/unit/test-staged-review-input-coverage.sh:35

OWASP LLM LLM03 Supply ChainOWASP Agentic ASI03 Identity and Privilege AbuseOWASP Agentic ASI04 Agentic Supply Chain VulnerabilitiesMITRE ATLAS AML.T0055 Unsecured CredentialsMITRE ATLAS AML.T0083 Credentials from AI Agent Configuration

environment = {k: v for k, v in os.environ.items() if not k.startswith('GIT_')}

What this means: Reads files or environment holding keys, tokens or sessions.

The rule's general advice: Confirm the tool needs this; otherwise do not install. (see the review note for this case)

Known false alarm

Test fixture in the plugin's own test suite: it creates or checks fake .env, settings or memory files under a temporary HOME, or filters the test process's own environment. Nothing reads real credentials or sends them anywhere.

Reviewed by Claude (AI-assisted review), 2026-11-21

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-staged-review-input-coverage.sh:21

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for tree in ('.claude/skills', 'skills', 'codex-skills'):

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-tangle-execution-boundary.sh:257

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

mkdir -p "$POLICY_CODEX_HOME/tmp" "$POLICY_HOME/.ssh" "$POLICY_HOME/.claude" \

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-task-resume-admission.sh:18

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

entries = [root / '.claude/skills/skill-task-management-v2/SKILL.md',

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-v8.41.0-feature-adoption.sh:153

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

AGENTS_DIR="$PLUGIN_ROOT/.claude/agents"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-validate-release-skill-registration.sh:86

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

shutil.copytree(fixture / 'skills', fixture / '.claude/skills')

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-validate-release-tag-scope.sh:93

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

for directory in ['.claude-plugin', '.codex-plugin', '.cursor-plugin', '.factory-plugin', 'commands', 'skills/flow-fixture', '.claude/skills/flow-fixture', 'codex-skills/flow-fixture']:

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-windows-doctor-compat.sh:219

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

if assert_contains "$doctor_generated" 'find "${HOME}/.claude/plugins"' "generated command searches Claude plugin installs" &&

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-workflow-meta-contracts.sh:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

CLAUDE_UI_UX_DESIGN="$PROJECT_ROOT/.claude/skills/skill-ui-ux-design/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-workflow-method-contracts.sh:10

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

ARCH="$PROJECT_ROOT/.claude/skills/skill-architecture/SKILL.md"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

tests/unit/test-workflow-result.py:94

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

path = ROOT / ('skills' if generated else '.claude/skills') / name / 'SKILL.md'

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

vendors/ui-ux-pro-max-skill/.claude-plugin/plugin.json:20

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"skills": "./.claude/skills/"

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

install-dir-access: Reads or writes Claude's install folders low

vendors/ui-ux-pro-max-skill/src/ui-ux-pro-max/templates/platforms/claude.json:6

MITRE ATLAS AML.T0081 Modify AI Agent Configuration

"root": ".claude",

What this means: Uses ~/.claude/agents, skills, commands or plugins (or a project .claude/ folder). Normal for installers, but it changes what Claude loads.

What you can do: Check which files it adds there and that you want them.

Dispute this result

Maintainer and think this is wrong? Open a dispute on GitHub with the target id plugin-nyldn-claude-octopus-2be7e9.