plan-technical-delivery
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources such as feature tickets and architecture specifications to create delivery plans.\n
- Ingestion points: The skill reads
feature-ticket.json,adr-spec.json, and various API contract schemas.\n - Boundary markers: There are no explicit text delimiters defined for the ingested data; however, the skill mandates that all contracts be validated against machine-readable JSON schemas.\n
- Capability inventory: The skill possesses the ability to
execute_command,write_file, andcreate_file, which are used for managing the planning documents and verifying implementation readiness.\n - Sanitization: Security is bolstered by strict schema validation for outputs and the requirement for Human-In-The-Loop (HITL) checkpoints for any slices marked for agent implementation.
Audit Metadata