optimise-claude
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill performs local analysis and editing of markdown files. It follows a defined workflow to maintain project instruction files and does not use obfuscation, external network requests, or privileged commands.- [COMMAND_EXECUTION]: The skill workflow includes a conditional step to run a local sync script to propagate changes. This is a documented administrative action within the project's development cycle.- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external instruction files. \n
- Ingestion points: CLAUDE.md, SKILL.md, and AGENTS.md files.\n
- Boundary markers: None detected.\n
- Capability inventory: File system write access for project instruction files.\n
- Sanitization: Not present, as the tool's purpose is to audit and refine instructions.
Audit Metadata