Skills
skills/levnikolaevich/claude-code-skills/ln-111-root-docs-creator/Gen Agent Trust Hub

ln-111-root-docs-creator

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local CLI commands including git worktree list, codex --version, and gemini --version to detect the presence and status of development tools for the tools_config.md file.\n- [COMMAND_EXECUTION]: The skill invokes MCP tool methods such as list_teams, ref_search_documentation, and resolve-library-id to identify and document active service providers in the environment.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting external data (project names and descriptions) and interpolating it into critical agent entry-point files like CLAUDE.md.\n
  • Ingestion points: SKILL.md (receives contextStore from an external coordinator).\n
  • Boundary markers: Absent; untrusted strings are directly interpolated into Markdown templates without delimiters.\n
  • Capability inventory: File system write access for documentation files, CLI execution for tool probing, and MCP tool invocation.\n
  • Sanitization: Absent; input data is used as-is without validation or escaping.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 01:44 PM