the-honoured-one
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a multi-phase protocol (Audit, Read, Orient, Gate) that requires the agent to identify and read all relevant files in a codebase before proposing or executing changes. This systematic reading of external files increases the exposure to indirect prompt injection if the ingested files contain malicious instructions.
- Ingestion points: Local source code, configuration files, and documentation identified during the "Phase 1 — Context Audit" (e.g., in
SKILL.md). - Boundary markers: The instructions do not specify the use of delimiters or specific warnings to ignore instructions embedded within the codebase files.
- Capability inventory: The protocol utilizes file reading capabilities to populate the agent's context session.
- Sanitization: There is no mention of sanitizing or filtering content read from the files before it is processed by the AI model.
Audit Metadata