project-features

Pass

Audited by Gen Agent Trust Hub on Nov 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process untrusted data from GitHub issues, including titles, bodies, and specific Markdown fields like ### web-feature-id. This data influences decision-making for triaging and automated workflows.
  • Ingestion points: Issue content from GitHub (titles, descriptions, and custom template fields) as described in SKILL.md.
  • Boundary markers: The process relies on specific Markdown headers (e.g., ### web-feature-id) and mentions a strict parser (extractFeatureIds) in the underlying project code.
  • Capability inventory: The agent is authorized to modify the repository by posting comments, editing issue bodies, changing labels, setting statuses, and closing issues. It also executes local shell commands.
  • Sanitization: The skill mentions that malformed identifiers fail silently in the parser, but the agent's actions based on the issue content require human confirmation as a safeguard.
  • [COMMAND_EXECUTION]: The skill directs the agent to execute shell commands to search the codebase and perform synchronization tasks.
  • Evidence: Instructions include running git grep to find feature IDs and node src/ci/generate-feature-to-groups.ts to register pending features in SKILL.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 10, 2026, 09:41 AM