Caveat verdict

skill-compass

88
🟢 Trusted
No high-risk patterns surfaced by the deep scan — automated capability review, not behavioral proof.

Reads files AND uses eval/Function

Evaluates skill quality and tracks usage by scanning local SKILL.md files and writing state to local JSON files; all operations are local file reads/writes with optional statusLine HUD configuration in ~/.claude/settings.json.

Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.

0
security
60
transparency
100
maintenance

What it does

These are capability combinations: each listed behavior occurs in the skill, but Caveat detects co-occurrence — it does not verify that one flows into another. Read the code to confirm a live chain.

Capability combination critical

Reads files AND uses eval/Function — may execute locally stored payloads

LLM05 · LLM06 · ASI05

Capability combination high

Accesses credentials AND writes files — may persist stolen credentials locally

LLM02 · LLM06 · ASI03

Capability combination high

Accesses agent memory AND writes files — may modify agent behavior persistently

LLM01 · LLM06 · ASI01 · ASI06

Capability combination medium

Both reads and writes files — verify scope is limited to intended directories

LLM06 · ASI02

Permission integrity

Performs file operations but does not declare file-accessing binaries

file_read+write

Code accesses API keys/tokens but declares no environment variables

credential_access

Accesses agent memory/configuration files

agent_memory

Findings (12)

Pattern match high

Writes to SKILL.md — self-modifying skill

commands/eval-improve.md · prose · downgraded · SKILL.md, refresh the self-write lock: update `.skill-compass/.write

Pattern match high

Accesses OpenClaw config/secrets directly

commands/setup.md · code · ~/.openclaw/openclaw.json

Pattern match high

Uses eval() — can execute arbitrary code

commands/skill-update.md · prose · downgraded · eval (

Pattern match high

Pipe to sh — executes piped content as shell commands

lib/patterns.js · prose · downgraded · |sh

Pattern match medium

Instructs covert action — may act without user awareness

SKILL.md · prose · downgraded · silently

Pattern match medium

Uses exec() — may execute shell commands

hooks/scripts/eval-gate.js · prose · downgraded · exec(

Pattern match medium

References child_process — can spawn system processes

lib/update-checker.js · prose · downgraded · child_process

Pattern match medium

References sudo — requests elevated privileges

prompts/d3-security.md · prose · downgraded · sudo

Pattern match low

Popular HTTP library — network access

lib/audit-chain.js · prose · downgraded · got

Pattern match low

References webhook/callback URL

lib/patterns.js · prose · downgraded · CALLBACK_URL

Pattern match low

Base64 encoding/decoding

lib/security-validator.js · prose · downgraded · base64-encode

Pattern match low

References tunneling service

shared/threat-signatures.yaml · prose · downgraded · ngrok

Why the tier is capped

Execution sink present in raw bytes (Hard Floor: class A/B/D/F). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.

Permissions & capabilities

Requires 1 system binary.

file_readfile_writecredential_accessagent_memorydynamic_eval
Check another skill Browse the registry Auditing your own skills or configs? Use the API