pr-description-writer

Warn

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/pr_body_lint.py dynamically loads the risk_classifier.py module at runtime using importlib.util and exec_module on a path computed relative to the script file. This dynamic loading from computed paths matches patterns for runtime code execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from pull request bodies and diff outputs, which could be exploited to influence agent behavior through indirect prompt injection.
  • Ingestion points: The skill fetches external pull request bodies using gh pr view and code changes using git diff.
  • Boundary markers: Absent. There are no delimiters or specific instructions for the agent to treat the ingested PR content as untrusted data.
  • Capability inventory: The skill can create and modify pull requests on the host behalf using the gh pr create and gh pr edit commands.
  • Sanitization: Absent. While scripts/pr_body_lint.py scans for secrets, it does not validate or sanitize inputs for malicious instructions or prompt injection patterns.
  • [CREDENTIALS_UNSAFE]: The file assets/sample_pr_body_flawed.md contains a hardcoded bearer token (Authorization: Bearer eyJhbGci...). Although this is a dummy value provided in a sample file for testing linter functionality, it matches the signature of hardcoded credentials.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 11, 2026, 03:18 AM