Caveat verdict
skill-pilot
Adaptive skill scheduling engine probing network for latency/DNS optimization and reading/writing local config files; its own security notes explicitly state it does not exfiltrate secrets.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
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.
Findings (10)
Recursive delete from root or home — destructive command
SECURITY_DECLARATION.md · code · rm -rf /
Accesses sensitive system files
SECURITY_DECLARATION.md · code · /etc/passwd
subprocess execution — runs system commands from Python
SECURITY_DECLARATION.md · code · subprocess.run(
Pipe to python — executes piped content as Python code
RELEASE.md · prose · downgraded · | Python
Writes to SKILL.md — self-modifying skill
scripts/init_skill.py · prose · downgraded · SKILL.md").write
Attempts to modify skills or system prompts
scripts/init_skill.py · prose · downgraded · Edit SKILL.md
subprocess with shell=True — command injection vector
SECURITY_FIX.md · code · subprocess.run(cmd, shell=True
os.system/popen — direct OS command execution
scripts/test.py · prose · downgraded · os.system(
importlib.import_module — dynamic module loading
scripts/engine.py · prose · downgraded · importlib.import_module(
Python os.getenv — reads environment variable
scripts/environment.py · prose · downgraded · os.getenv(
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.