write-tech-radar
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to summarize external technology context, trade-offs, and vendor claims, which introduces an attack surface for indirect prompt injection. A malicious primary source or vendor documentation could attempt to influence the agent's output or recommendations.
- Ingestion points: Technical documentation, vendor claims, and technology context read from the project via
read_fileandsearch_code(referenced in SKILL.md). - Boundary markers: The skill does not define explicit delimiters to isolate external content from instructions, though it does include safety headers.
- Capability inventory: The skill utilizes
write_file,edit_file, andcreate_fileto produce JSON contracts and Markdown artifacts (referenced in SKILL.md). - Sanitization: The instructions explicitly mitigate risk by requiring validation against primary sources and treating vendor marketing as untrusted (ASI01 and ASI04 guardrails).
Audit Metadata