company-creator

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external git repositories to generate agent descriptions and organizational workflows.
  • Ingestion points: The skill clones repositories and scans files like README.md, package.json, and source code to identify natural agent roles (as detailed in references/from-repo-guide.md).
  • Boundary markers: No specific delimiters or safety warnings for the LLM regarding embedded instructions are specified in the generation instructions.
  • Capability inventory: The skill has the capability to write complex directory structures and multiple markdown files to the local file system. It also performs network operations using git tools.
  • Sanitization: There is no mention of automated sanitization of ingested content. However, the skill enforces a mandatory AskUserQuestion interview step before file generation, requiring the user to confirm the proposed agents, roles, and workflows.
  • [EXTERNAL_DOWNLOADS]: The skill performs network operations to clone remote repositories and fetch commit metadata via git ls-remote to support its 'Company From a Repo' functionality.
  • [COMMAND_EXECUTION]: The skill uses shell commands, specifically git, to interact with external codebases for analysis and to resolve specific commit SHAs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 08:48 AM
Security Audit — agent-trust-hub — company-creator