vertical-slice-delivery

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow that processes user-provided task descriptions and markdown plan files, creating an indirect prompt injection surface common to orchestration skills.
  • Ingestion points: Task requests and markdown plan files stored in the plans/ directory.
  • Boundary markers: Absent; the skill relies on role-based context separation instead of explicit delimiters.
  • Capability inventory: Git command execution (status, log, add, commit) and file writing within the plans/ directory.
  • Sanitization: Absent; behavior is managed via specialized sub-agent prompts.
  • [SAFE]: The skill's primary operations involve standard software development tasks such as code implementation, formatting, and repository management.
  • [SAFE]: Git commands are used solely for local workspace management and include appropriate safety measures, such as clearing the staging area before adding slice-specific files.
  • [SAFE]: The orchestrator enforces a human-in-the-loop permission check before enabling automatic commits, ensuring the user retains control over the repository state.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 09:49 PM
Security Audit — agent-trust-hub — vertical-slice-delivery