project-init

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill executes inline JavaScript via node -e to interact with local repository-based scripts like session-init-helpers.cjs and agent-files-state.cjs. This is used for assessing project state and validating configurations.
  • [COMMAND_EXECUTION]: The skill uses Git commands (git mv, git rm) to manage file naming conventions and handle duplicate files during the normalization of project documentation.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific documentation and configuration files, creating a surface for indirect prompt injection.
  • Ingestion points: Reads docs/project-config.json, docs/project-reference/*.md, CLAUDE.md, and AGENTS.md.
  • Boundary markers: Includes instructions to respect user-authored content and use smart-merge logic for updates.
  • Capability inventory: Spawns subprocesses for Node.js and Git, and utilizes sub-agent spawning for task parallelization.
  • Sanitization: No explicit sanitization or validation of the text content within these files is performed before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 03:12 PM
Security Audit — agent-trust-hub — project-init