ClawAudit verdict
deep-research-engine
Deep research workflow skill that uses web_search and web_fetch tools to conduct multi-source research and generate reports; package_install capability is for declared dependencies (deepagents, tavily-python, langchain-anthropic) with no exfiltration or injection patterns.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Permission integrity
package_install
Findings (3)
Possible hardcoded credential
SKILL.md · code · API_KEY="your_key
Python os.getenv — reads environment variable
backend/agent.py · prose · downgraded · os.getenv(
Python httpx request — network access
backend/agent.py · prose · downgraded · httpx.get(
Permissions & capabilities
No declared permissions — minimal attack surface.
package_install Is this flag fair?
Thanks — recorded.