skills/soderlind/skills/wp-prepare/Gen Agent Trust Hub

wp-prepare

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches agent skills from the official WordPress GitHub organization using npx skills add. These resources provide the agent with specialized capabilities for WordPress plugin and block development.
  • [EXTERNAL_DOWNLOADS]: Downloads coding and security instructions from a public repository (github/awesome-copilot) to local project storage at .github/instructions/. These files help ensure that the agent follows WordPress-specific best practices.
  • [COMMAND_EXECUTION]: Automates project setup using shell commands like npm init, git init, and composer require. Command execution in scripts/plan_setup.mjs is handled with hardcoded templates to mitigate the risk of shell injection from user-provided metadata.
  • [INDIRECT_PROMPT_INJECTION]: The skill establishes an indirect prompt injection surface by importing an external markdown instruction file into the project context.
  • Ingestion points: The wordpress.instructions.md file is downloaded via curl in SKILL.md (Phase 6b) and scripts/plan_setup.mjs.
  • Boundary markers: None; the downloaded content is integrated directly as a set of rules for the AI agent to follow.
  • Capability inventory: The agent possesses capabilities to execute shell commands (npm, git, composer), perform file system operations, and manage dependencies within the repository.
  • Sanitization: The skill does not sanitize the contents of the downloaded instruction file before placing it in the project directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 12:10 PM
Security Audit — agent-trust-hub — wp-prepare