skills/mkrlabs/specflow/github-pr/Gen Agent Trust Hub

github-pr

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes common development CLI tools including git, gh (GitHub CLI), and gcloud (Google Cloud SDK). These commands are used to manage branches, create pull requests, and retrieve infrastructure metadata for preview environments. It also uses git push --no-verify, which bypasses local git hooks, a common practice in automated workflows.
  • [DATA_EXFILTRATION]: Local branch metadata, commit history, and code differences are transmitted to GitHub through git push and gh pr create. This data transfer is required for the skill's primary function of PR creation.
  • [PROMPT_INJECTION]: The skill processes potentially untrusted content from commit messages, local specification files, and GitHub PR comments to construct PR descriptions and report status to the user, creating a surface for indirect prompt injection.
  • Ingestion points: Reads data from specs/<branch-name>/spec.md, git log, and GitHub PR comments via gh pr view.
  • Boundary markers: None identified; untrusted content is interpolated directly into command arguments and reporting templates.
  • Capability inventory: The skill has the capability to push to remote repositories, create/edit Pull Requests, and query status for GCP Cloud Run services.
  • Sanitization: PR titles are automatically truncated to 70 characters. The skill also provides a robust verification method for preview URLs by querying gcloud directly as a fallback to untrusted PR comments.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 03:08 PM
Security Audit — agent-trust-hub — github-pr