decision-record
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill operates as a structural template for organizing user-provided information. It does not contain executable code, remote script downloads, or credential-harvesting patterns.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text regarding technical decisions and context to generate documentation. While this constitutes an ingestion surface for untrusted data, the skill includes procedural instructions to maintain neutral facts and strictly use provided figures, which mitigates the risk of the agent following instructions embedded within the user data.
- Ingestion points: User descriptions of decisions, context, and options are processed to generate the ADR (SKILL.md).
- Boundary markers: None explicitly defined in the instructions.
- Capability inventory: The skill can interface with integrated notes tools or the vendor's connector to save Markdown output (SKILL.md).
- Sanitization: The skill instructs the agent to use only provided facts and figures and maintain neutrality, serving as a procedural safeguard.
- [COMMAND_EXECUTION]: The skill describes the capability to save generated records using connected notes tools or a vendor-specific connector (Scratchpad). This is a functional capability consistent with the primary purpose of a documentation skill.
Audit Metadata