skills/sones3/matt-skills/to-issues/Gen Agent Trust Hub

to-issues

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data, which creates a surface for indirect prompt injection attacks.
  • Ingestion points: The agent is instructed to fetch and read issue bodies and comments from an issue tracker, and process user-provided documents like PRDs or plans (SKILL.md, Step 1).
  • Boundary markers: There are no explicit delimiters or specific instructions provided to the agent to treat ingested data as untrusted or to ignore embedded commands.
  • Capability inventory: The agent has permissions to read the local codebase and write (publish) new issues to a project's issue tracker.
  • Sanitization: The instructions do not specify any sanitization for the input data. However, the skill includes a mandatory human-in-the-loop (HITL) review step (Step 4: Quiz the user) which provides a checkpoint for the user to verify the agent's proposed actions before any data is published.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 06:52 PM