git-project

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub (issue titles, bodies, and project item content) which provides an attack surface for indirect prompt injection.
  • Ingestion points: Data is ingested from external sources via gh issue view, gh project item-list, and various GraphQL queries documented in references/cli-and-graphql.md and references/sub-issues.md.
  • Boundary markers: The playbooks and reference commands do not include explicit delimiters or instructions for the agent to ignore potentially malicious instructions embedded within the issue data being processed.
  • Capability inventory: The skill utilizes powerful write-capable commands, including gh issue create/edit, gh project item-edit, and gh api POST/PATCH mutations, which could be misused if the agent obeys instructions hidden in the ingested metadata.
  • Sanitization: No sanitization or validation logic is prescribed for handling the external data before it is used in repository operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 03:55 PM
Security Audit — agent-trust-hub — git-project