performance
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external performance data that could potentially contain malicious instructions.
- Ingestion points: Reads performance traces stored in .scratchpad/ and workload definitions (references/perf-issue.md, references/hillclimb.md).
- Boundary markers: No explicit instructions are provided to the agent to treat trace data as untrusted or to ignore instructions embedded within it.
- Capability inventory: The skill can modify the repository (git), create pull requests (gh), and execute arbitrary benchmark harnesses.
- Sanitization: No validation or sanitization of the content within ingested traces or benchmark artifacts is performed.
- [COMMAND_EXECUTION]: The skill executes shell commands to manage code and interface with GitHub.
- Evidence: Uses git add, git commit, and gh pr create to manage the lifecycle of performance improvements (references/hillclimb.md, references/perf-issue.md).
Audit Metadata