idea-memo-writer
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions to execute a local Python script using
uv run python. This script (scripts/run.py) performs deterministic assembly of research reports from local workspace files. - [DATA_EXPOSURE_EXFILTRATION]: No network operations or external data transmissions were detected. The script reads from local paths like
output/trace/IDEA_SHORTLIST.jsonland writes to theoutput/directory within the specified workspace. - [REMOTE_CODE_EXECUTION]: All code is contained within the skill folder or referenced from local
toolingmodules within the agent's environment. There are no downloads of external scripts or packages from untrusted sources. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided research artifacts (JSONL and CSV files). While these are ingestion points for external data, the script uses structured parsing and deterministic rendering logic, which limits the surface for prompt injection attacks. No unsafe interpolation into agent prompts was identified.
Audit Metadata