evidence-driven-agent-rules
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it reads and analyzes failure logs authored by users.
- Ingestion points: The
promoteworkflow reads multiple files matchingdocs/reflection-log/[0-9]*.mdto identify recurring patterns. - Boundary markers: No specific boundary markers or 'ignore' instructions are used when interpolating log content into the agent's context during rule generation.
- Capability inventory: The skill can execute
grepvia the shell and has the authority to write toAGENTS.md, git hooks, and CI configurations. - Sanitization: Content from the logs is not sanitized; however, the skill explicitly instructs the agent to ask for confirmation before writing any changes.
- [COMMAND_EXECUTION]: The skill utilizes shell commands, primarily
grep, for pattern matching across log files. It also includes a developer utility script,evals/run-static-checks.sh, which executes standard development tools likegit,python3, andreadlinkto validate the skill structure and schema compliance.
Audit Metadata