add-expert

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires the agent to execute the shell command bun render-cards within the packages/docs directory. This is a local script execution used to generate assets based on the updated project data.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data to populate source code, which introduces a potential surface for indirect prompt injection.
  • Ingestion points: User-provided expert details including the name, website, and a JSX description field in packages/promo-pages/src/components/experts/experts-data.tsx.
  • Boundary markers: Absent. The skill uses a template for direct interpolation of user data into code blocks without specifying delimiters or instructions to ignore embedded commands.
  • Capability inventory: The agent has the capability to write to the local file system and execute shell commands (bun).
  • Sanitization: Absent. The instructions explicitly permit the use of HTML tags like <a> within the description, which increases the risk that malicious instructions could be embedded in the data and later processed by the agent or other tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:01 PM
Security Audit — agent-trust-hub — add-expert