write-implementation-plans
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, such as issue tracker content, pull-request descriptions, and documentation retrieved from the web, to generate implementation plans.\n
- Ingestion points: Untrusted data enters the agent context through requirements documents, issue descriptions, pull-request bodies, and retrieved web content, as specified in
SKILL.mdandreferences/enterprise-operation.md.\n - Boundary markers: The skill contains explicit instructions in
references/enterprise-operation.mdto treat external data as untrusted and to ignore any embedded instructions that attempt to escalate privileges or broaden scope.\n - Capability inventory: The skill utilizes existing repository tools and commands for code inspection and contract verification, focusing on plan generation rather than execution, as described in
SKILL.mdandreferences/verification-and-evidence.md.\n - Sanitization: Instructions mandate the use of least-privilege credentials and strictly prohibit the inclusion of secrets in logs, artifacts, or skill-related files.\n- [EXTERNAL_DOWNLOADS]: The skill references documentation and industry standards from well-known and reputable external organizations for information gathering.\n
- Evidence:
references/source-index.mdprovides links to official guidelines and design principles from Google, Microsoft, NASA, the GAO, and the IETF (RFC Editor) to support implementation decisions and verification procedures.
Audit Metadata