understand-knowledge

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes bundled Python scripts (parse-knowledge-base.py, merge-knowledge-graph.py) to process wiki data.
  • [COMMAND_EXECUTION]: System commands are used for maintenance and metadata retrieval.
  • git rev-parse HEAD is called in merge-knowledge-graph.py to record the project state.
  • rm -rf is used in SKILL.md (Phase 5) to clean up intermediate analysis files.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted markdown content.
  • Ingestion points: parse-knowledge-base.py reads markdown files from a user-specified directory.
  • Boundary markers: Absent. Instructions in SKILL.md (Phase 3) pass raw content from knowledgeMeta directly to subagents without delimiters or instructions to ignore embedded commands.
  • Capability inventory: The skill has file-system access (read/write/delete) within the target directory and execution capabilities for local scripts.
  • Sanitization: No sanitization or validation of the markdown content is performed before it is passed to LLM subagents.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 06:21 PM
Security Audit — agent-trust-hub — understand-knowledge