The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown and does not contain any executable scripts, binary files, or configuration that triggers command execution.
[PROMPT_INJECTION]: The skill processes untrusted external data, specifically issue bodies and pull request comments, which introduces a surface for indirect prompt injection.
Ingestion points: Workflow steps 'Intake' (Step 1) and 'Review loop' (Step 4) read external issue descriptions and PR feedback.
Boundary markers: No explicit delimiter instructions are provided for the intake of external text.
Capability inventory: The skill orchestrates sub-skills like closed-loop-delivery and git-ship, which typically involve file system access and git operations.
Sanitization: No explicit sanitization or filtering of external data is defined.
Mitigation: The risk is mitigated by explicit 'Human Gates' requiring manual confirmation for destructive operations, security changes, and production deployments, as well as a strict state machine that requires verifiable evidence for completion.

acceptance-orchestrator