grimoire-plan
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions guide the agent to start a local HTTP server for previewing plans. It attempts to execute
python -m http.serverorpy -m http.serverbound to the loopback address127.0.0.1as a background process to serve files locally. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from repository code, Architecture Decision Records (ADRs), and tickets to generate implementation plans.
- Ingestion points: The agent reads repository source code, local knowledge files, and ticket descriptions to build context.
- Boundary markers: There are no specific instructions to use delimiters or ignore embedded commands within the ingested content.
- Capability inventory: The skill has the ability to write files to the
.grimoire/plans/directory and initiate local network services. - Sanitization: The included
assets/plan-viewer.htmlutilizesDOMPurifyto sanitize rendered markdown, which effectively mitigates XSS risks during the preview process. - [EXTERNAL_DOWNLOADS]: The plan viewer tool (
assets/plan-viewer.html) fetches external JavaScript dependencies from the JSDelivr CDN. These resources includemarked.jsfor markdown rendering,DOMPurifyfor security sanitization, andmermaid.jsfor diagram generation.
Audit Metadata