agile-v-sop-adapter
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill manages an attack surface for indirect prompt injection by processing organizational binding files (
.agile-v/SOP_BINDING.yaml) that dictate agent actions and verification steps.\n - Ingestion points: The agent and the
validate.pyscript read external YAML configurations stored in the project repository.\n - Boundary markers: The instructions mandate referencing only clause IDs and forbid embedding normative SOP text, reducing the volume of untrusted content processed by the LLM.\n
- Capability inventory: The skill is permitted to append evidence to
.agile-v/APPROVALS.mdand.agile-v/CHECKPOINTS.md. The validator script does not possess network or unauthorized file system capabilities.\n - Sanitization: The validation utility utilizes
yaml.safe_load()to ensure secure parsing of YAML files, preventing arbitrary code execution during deserialization.
Audit Metadata