read-all-adrs
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local file system (ADR files) without implementing boundary markers or content sanitization, creating a surface for indirect prompt injection.
- Ingestion points: The agent is instructed to read every
.mdfile in thedocs/adr/directory (SKILL.md). - Boundary markers: The instructions lack delimiters or warnings to the agent to ignore instructions embedded within the file content.
- Capability inventory: The skill is focused on file reading and context summarization (SKILL.md).
- Sanitization: There are no mechanisms described to validate, filter, or escape the content of the ADR files before they are read into the agent's context.
Audit Metadata