solo-deploy

Pass

Audited by Gen Agent Trust Hub on Apr 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform extensive operations including git management, repository creation via the GitHub CLI, and execution of platform-specific deployment tools (Vercel, Wrangler, Supabase, SST). It includes logic to handle potential package installations if tools are missing.
  • [EXTERNAL_DOWNLOADS]: In autonomous mode, the skill may download and install missing platform CLI tools from well-known sources using npm or Homebrew (e.g., Vercel, Wrangler, Supabase).
  • [PROMPT_INJECTION]: The skill uses dynamic context injection to insert real-time git state information into the prompt during load. The commands used (git branch, status, log) are benign and serve to provide context for the deployment.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is instructed to prioritize deployment steps found within the project's own documentation.
  • Ingestion points: Project configuration and documentation files, including CLAUDE.md, package.json, and specifically docs/plan/*/plan.md.
  • Boundary markers: No specific delimiters or safety warnings are implemented to isolate or sanitize instructions extracted from these files.
  • Capability inventory: The agent has broad shell access, file system modification rights, and network connectivity via the Bash tool.
  • Sanitization: No validation or escaping of the ingested plan content is performed prior to the execution of the described deployment tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 5, 2026, 09:51 AM