organise-agent-skills

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions to identify and delete obsolete skill directories within the ~/.gemini/config/skills/ path as part of the pruning workflow.
  • [DATA_EXFILTRATION]: The agent is directed to read and audit sensitive configuration and metadata files, including skills.json, GEMINI.md, and AGENTS.md, which contain information about the agent's capabilities and environment.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and acts upon the content of local configuration files which could be manipulated by other tools or skills. Evidence Chain: 1. Ingestion points: skills.json, GEMINI.md, AGENTS.md. 2. Boundary markers: Absent. 3. Capability inventory: File mutation (write) and directory deletion (rm). 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 01:03 PM
Security Audit — agent-trust-hub — organise-agent-skills