skills/zernie/vigiles/dogfood-cli/Gen Agent Trust Hub

dogfood-cli

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze external code and plugins for bug hunting. This creates a surface for indirect prompt injection, where malicious instructions embedded in the audited files (such as code comments or docstrings) could attempt to influence the agent's behavior during the 'find and fix' loop. \n
  • Ingestion points: Source code files of the vigiles CLI and external plugins processed during the dogfooding loop. \n
  • Boundary markers: The skill lacks explicit boundary markers or instructions to ignore embedded commands, although it mandates 'source-tracing' findings to specific file lines. \n
  • Capability inventory: Extensive local capabilities including file system writes, Git repository management (commit, push, worktree), and execution of development toolchains. \n
  • Sanitization: No explicit sanitization of the content being audited is mentioned.\n- [COMMAND_EXECUTION]: The workflow relies on the execution of several CLI tools and scripts, including npm run build, vitest, prettier, and git. While standard for a developer role, these tools are the primary vector for any actions the agent might be misled into performing through the analysis of untrusted plugin code.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 02:32 PM
Security Audit — agent-trust-hub — dogfood-cli