do-agent-context-layer
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown documentation and organizational guidance. It does not contain any executable scripts, binary files, or tool definitions.
- [INDIRECT_PROMPT_INJECTION]: The skill outlines a methodology for how agents ingest repository context, which defines the surface area for potential indirect prompt injections.
- Ingestion points: Guidance is provided for structuring entry files (
AGENTS.md) and documentation trees (docs/) that are used by agents to build project mental models. - Boundary markers: The methodology encourages "Pointers over payload" and the use of directory catalogues (
index.md) to provide structure and scope, helping agents identify relevant context sections. - Capability inventory: The skill itself possesses no executable capabilities or system-level tools.
- Sanitization: Recommends structural validation through CI and suggests that truth-based invariants should be enforced by code tests rather than prose documentation.
Audit Metadata