workspace-state
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a conceptual and technical guide for implementing a workspace state harness, focusing on reliability and security in agent-filesystem interactions.
- [PROMPT_INJECTION]: Addresses the surface for Indirect Prompt Injection by defining strict state invariants. 1. Ingestion points: File reading tool definitions in SKILL.md. 2. Boundary markers: Implementation of hash-based verification in the edit tool logic. 3. Capability inventory: Support for file modification and shell command execution. 4. Sanitization: Recommendation for workspace root enforcement and canonicalization of paths.
Audit Metadata