Caveat verdict

subagent-context-compactor

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

A local context compression system that monitors session token usage and applies tiered summarization using local scripts; all operations are internal session management with no data sent to external services.

⚠ 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.

0
security
70
transparency
70
maintenance

Permission integrity

Installs packages at runtime — transitive dependencies are not auditable

package_install

Findings (7)

Pattern match high

HTTP request to bare IP address — common in malicious payloads

DEPLOYMENT.md · code · http://127.0.0.1

Pattern match high

Uses exec() — may execute shell commands

OPENCLAW_INTEGRATION.md · code · exec(

Pattern match high

Possible hardcoded credential

OPENCLAW_INTEGRATION.md · prose · downgraded · API_KEY="your-secret-key

Pattern match medium

os.system/popen — direct OS command execution

api_server.py · prose · downgraded · os.popen(

Pattern match medium

pip3 install — installs Python packages at runtime

DEPLOYMENT.md · code · pip3 install

Pattern match medium

Python os.environ.get — reads environment variable

DEPLOYMENT.md · code · os.environ.get(

Pattern match medium

subprocess execution — runs system commands from Python

integration.py · prose · downgraded · subprocess.run(

Why the tier is capped

Execution sink present in raw bytes (Hard Floor: class B/D/F). 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.

network_inpackage_install

Is this flag fair?

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