mindmap

Pass

Audited by Gen Agent Trust Hub on Sep 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests content from external URLs via WebFetch and local files via Read to generate mindmap structures.
  • Ingestion points: External content from URLs and local file system content (SKILL.md).
  • Boundary markers: The multi-agent workflow in references/judge-panel.md uses specific instructions to condense content into short phrases, but lacks explicit delimiters or sandbox warnings for the untrusted data.
  • Capability inventory: Access to tool-based file writing (Write) and shell command execution (Bash).
  • Sanitization: The transformation from long-form text to short mindmap nodes acts as a structural filter, reducing the likelihood of instruction persistence.
  • [EXTERNAL_DOWNLOADS]: The scripts/render.sh utility invokes npx --yes markmap-cli, which downloads the Markmap package from the npm registry at runtime.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute a local shell script (render.sh) for HTML generation and a Node.js script (degrade-rich.mjs) for markdown processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 22, 2026, 02:06 PM
Security Audit — agent-trust-hub — mindmap