okf-knowledge-base
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process external Markdown files and YAML frontmatter within a directory tree (bundles). This process establishes a surface for indirect prompt injection, particularly as the agent is directed to trust the bundle content when conflicts with assumptions arise.
- Ingestion points: The agent is instructed to read all
.mdfiles within a bundle directory tree (SKILL.md). - Boundary markers: The instructions specify using structured YAML frontmatter and standard Markdown links as delimiters for content and metadata.
- Capability inventory: The agent relies on the OpenKnowledge MCP server and the
okfplugin to provide conformance feedback and manage bundle files. - Sanitization: While the
okfplugin performs structural audits and schema validation, the skill does not detail specific sanitization for natural language instructions that might be embedded within the Markdown content. - [EXTERNAL_DOWNLOADS]: The skill and associated metadata reference external repositories for technical specifications and project resources. These references target the official GoogleCloudPlatform GitHub organization and the author's own development repository (Inkeep).
Audit Metadata