automate-me
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill mines conversation history to generate behavioral rules, which creates a surface for indirect prompt injection if historical transcripts contain adversarial content.\n
- Ingestion points:
agent-transcripts/directory within the active workspace.\n - Boundary markers: None specified for the transcript content mining.\n
- Capability inventory: File creation via
create-skilland repository modification viagit.\n - Sanitization: Employs cross-slice verification to identify high-confidence patterns, acting as a heuristic filter.\n- [DATA_EXFILTRATION]: The skill analyzes sensitive conversation history. It implements a privacy guardrail by restricting data access to the active workspace's
agent-transcripts/directory and explicitly forbidding searches across other project directories.\n- [COMMAND_EXECUTION]: The skill utilizesgitcommands (git log,git worktree,git commit) to manage skill versions and stage changes for human review via pull requests.
Audit Metadata