feature-delivery
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is entirely composed of markdown-based process guidelines and templates for software delivery management. It does not contain any executable scripts, shell commands, or perform any network requests.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze external repository content including code, tests, and architecture. 1. Ingestion points: Repository files and environment states are inspected during the discovery and planning phases. 2. Boundary markers: The process uses structured documentation templates and explicit definitions of non-goals and unresolved decisions. 3. Capability inventory: The agent's capabilities are restricted to file system reading and reporting; mutations such as code merges, build publication, or deployment require explicit user authorization. 4. Sanitization: Instructions explicitly mandate the exclusion of credentials, private product data, and production data from all output reports and specifications.
Audit Metadata