open-knowledge-discovery

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references the installation of the @inkeep/open-knowledge package from the official npm registry and downloading the OpenKnowledge Desktop application from the vendor's GitHub releases page.- [COMMAND_EXECUTION]: The instructions guide users to execute commands such as ok init, ok cowork, and ok open to initialize projects, set up collaboration tools, and preview markdown files.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to handle markdown (.md) and MDX files, which are external data sources that could potentially contain embedded instructions.
  • Ingestion points: The agent processes markdown files when the user asks to open or preview them (SKILL.md).
  • Boundary markers: The discovery skill explicitly states that the runtime safety contract and grounding rules are handled by a separate project-local skill installed during the initialization process.
  • Capability inventory: The skill utilizes the ok CLI for initialization and the preview_url MCP tool for rendering file previews.
  • Sanitization: Safety protocols for reading and writing markdown content are managed by the project-local runtime skill rather than this discovery skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 08:09 PM
Security Audit — agent-trust-hub — open-knowledge-discovery