finishing-a-development-branch
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or hidden obfuscation were detected. The skill instructions follow security best practices by explicitly prohibiting automated mutations like force-pushes or auto-stashing and requiring granular human confirmation for resource deletion.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it reads repository state and metadata which could be influenced by untrusted contributors. However, this is mitigated by: 1. Ingestion points: Reads repository root, HEAD, branch state, and worktree lists in Step 1. 2. Boundary markers: None explicitly defined in the provided instruction text. 3. Capability inventory: Includes git merge, push, and worktree/branch deletion capabilities. 4. Sanitization: Implements strict ownership-aware cleanup, mandatory fresh verification, and exact typed confirmation requirements for all deletions.
Audit Metadata