pi-planning-with-files

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes several shell and PowerShell scripts (init-session.sh, attest-plan.sh, check-complete.sh) to automate project management. These scripts handle user-supplied project names safely by applying slugification (removing special characters) and using literal path parameters, which effectively mitigates command injection risks.
  • [DATA_EXPOSURE]: The session-catchup.py script reads session history logs from local directories (~/.claude/projects/ or ~/.codex/sessions/) to help the agent recover context after a reset. This data processing is entirely local, and the script contains no network-capable code for exfiltrating this information.
  • [REMOTE_CODE_EXECUTION]: All executable logic is contained within the skill's local directory. There are no patterns involving the download and execution of remote scripts (e.g., curl | bash) or use of eval on untrusted external input.
  • [SAFE]: The skill follows security best practices, such as using file hashing for plan integrity (attestation) and providing clear templates. No prompt injections or obfuscation techniques were found.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 11:02 AM