vulture-sweep
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from GitHub issue trackers, creating a surface for indirect prompt injection attacks.
- Ingestion points: The skill retrieves issue metadata, titles, and bodies using
gh issue list --json number,title,labels,createdAt,updatedAt,body(SKILL.md, Phase 1). - Boundary markers: The instructions lack explicit boundary markers or delimiters when interpolating issue content into the agent's reasoning process.
- Capability inventory: The agent has the capability to perform repository searches (
gf,git log), close issues (gh issue close,gw gh issue batch), and create new issues (gh issue create). - Sanitization: There are no documented sanitization steps or instructions to ignore embedded commands within the issue text, meaning a malicious issue could potentially trick the agent into closing unrelated issues or creating misleading ones.
Audit Metadata