adr
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to read existing files in the repository's ADR directory and "Obey any 'do not' line" found, making the agent's behavior dependent on the content of these records.
- Ingestion points: The skill reads markdown files within the ADR directory (e.g.,
docs/adr/) to retrieve context before code modifications. - Boundary markers: No specific boundary markers or instruction-isolation techniques are implemented for the content read from these files.
- Capability inventory: The skill allows the agent to read and write markdown files; it does not possess capabilities for network exfiltration or shell execution.
- Sanitization: The skill does not describe any sanitization or validation of the text found in existing ADRs before the agent adopts it as policy.
- [NO_CODE]: The skill is comprised of markdown instructions and evaluation datasets (text and TSV files) without any executable scripts or binary files.
Audit Metadata