skills/basher83/lunar-claude/git-push/Gen Agent Trust Hub

git-push

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes output from local Git commands, which can theoretically contain untrusted data if an attacker controls remote branch names or repository URLs.
  • Ingestion points: The skill reads output from git remote get-url origin, git branch --show-current, and git rev-list (SKILL.md).
  • Boundary markers: No explicit delimiters are used to wrap the output of Git commands when they are processed by the agent.
  • Capability inventory: The skill is restricted to standard Git operations, including git fetch and git push.
  • Sanitization: The instructions do not specify any validation or sanitization steps for the strings returned by Git commands before they are evaluated for the workflow logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 04:53 PM
Security Audit — agent-trust-hub — git-push