day6-prd-submit
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill performs several shell operations using
gitandgh(GitHub CLI) to manage the local environment and project workflow. - Evidence: Execution of
git checkout,git pull,git add,git commit, andgit pushto manage document versions. It also usesgh pr createto automate pull request generation. - [DATA_EXFILTRATION]: The skill is designed to transmit local documentation to an external repository on GitHub.
- Evidence: The command
git push origin prd/{github-id}uploads the user's PRD file to a remote server. GitHub is recognized as a well-known service for this functionality. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local files which could potentially contain malicious instructions if the files are compromised or modified by external actors.
- Ingestion points: The skill reads the content of
{github-id}/PRD.mdfor formatting validation and scans the.claude/skills/directory to list available skills. - Boundary markers: Absent. The skill relies on specific Markdown header detection rather than secure delimiters for the ingested content.
- Capability inventory: File system writing, Git branch management, and remote repository interaction (push and PR creation).
- Sanitization: The skill implements a robust validation check for the
github-idinput, restricting it to alphanumeric characters and hyphens, which effectively prevents shell command injection through that parameter. - [EXTERNAL_DOWNLOADS]: The skill provides instructions for the user to download and install the GitHub CLI tool.
- Evidence: Suggests
brew install ghfor macOS users. The GitHub CLI is a recognized tool from a well-known service.
Audit Metadata