team-composer

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill orchestrates a multi-persona brainstorming session using structured instructions and persona definitions.
  • [PROMPT_INJECTION]: The skill contains logic-gate instructions (e.g., Phase 0 stop gates and Phase 3 lock-in checks) designed to enforce a specific workflow and prevent the agent from bypassing required domain experts. These are protective measures for output quality and alignment, not attempts to subvert AI safety filters.
  • [EXTERNAL_DOWNLOADS]: The documentation and evaluation files reference external resources and repositories for cross-skill integration. These include trusted organizations (GoogleChrome) and the author's own GitHub repositories (sorawit-w), primarily used for linking to related tools, assets, and documentation. These are documented neutrally and follow trust-scope rules.
  • [DATA_EXFILTRATION]: No patterns of unauthorized data access or network exfiltration were found. The skill processes user-provided project briefs solely for the purpose of identifying project signals and roles to drive a text-based discussion.
  • [COMMAND_EXECUTION]: No evidence of shell command execution or dangerous subprocess spawning was found. The skill's primary function is generating a synthesized text-based plan for the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 07:12 AM
Security Audit — agent-trust-hub — team-composer