openclaw-small-bugfix-sweep
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted data from external sources (GitHub issues and PRs), creating a surface for indirect prompt injection where malicious instructions in the data could attempt to influence the agent.
- Ingestion points: The skill reads GitHub issue bodies, PR descriptions, and comments using the
ghtool as specified in the 'Loop' section ofSKILL.md. - Boundary markers: There are no explicit markers or instructions defined to separate external data from the agent's system instructions.
- Capability inventory: The agent can modify local files and execute 'gates' (tests/build scripts) as described in
SKILL.md. However, sensitive actions like commits are gated by explicit human approval. - Sanitization: No sanitization or validation of the external content is performed before processing.
Audit Metadata