context-onboarding

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest data from external sources such as documentation pages, database tables, and APIs to build new AI skills. This creates a surface for indirect prompt injection, where malicious instructions embedded in the source data could influence the agent's behavior or the generated output.
  • Ingestion points: The skill reads content from external sources (files, URLs, warehouse tables via MCP) to populate evidence snippets and the project.json file.
  • Boundary markers: The skill uses a structured JSON format (project.json) for data storage, but does not explicitly define sanitization or boundary markers for the internal content of text snippets.
  • Capability inventory: The skill uses pnpm for exporting and scripts/profile_table.sql for database interactions.
  • Sanitization: While the skill correctly identifies and avoids storing credentials, it does not mention sanitizing external text for potential prompt injection attacks.
  • [COMMAND_EXECUTION]: The workflow requires the execution of local shell commands using pnpm (e.g., pnpm export:skill) to process the gathered data and generate the final skill artifacts. This is a standard development practice for this type of tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 11:47 AM
Security Audit — agent-trust-hub — context-onboarding