jmb-improve-codebase-architecture

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands to interact with the host environment and display its results.
  • Executes platform-specific commands (xdg-open, open, or start) to automatically open the generated HTML report in the user's browser.
  • Runs git log --oneline to analyze the project's commit history for identifying development hot spots.
  • [EXTERNAL_DOWNLOADS]: The generated architectural reports are configured to fetch resources from well-known external services at runtime.
  • Loads the Tailwind CSS framework from https://cdn.tailwindcss.com.
  • Loads the Mermaid diagramming library from https://cdn.jsdelivr.net/npm/mermaid@11/dist/mermaid.esm.min.mjs.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase data which could contain malicious instructions designed to influence the agent's architectural analysis.
  • Ingestion points: Reads the entire codebase, including source files, documentation (CONTEXT.md, ADRs), and git commit messages via sub-agents (SKILL.md).
  • Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded prompts within the scanned codebase.
  • Capability inventory: Performs file writes to the system temp directory, executes shell commands to open files, and calls external tools like jmb-grilling and jmb-domain-modeling (SKILL.md).
  • Sanitization: No explicit sanitization or filtering of codebase content is mentioned before it is processed or rendered into the HTML report.
  • [DYNAMIC_EXECUTION]: The reporting template utilizes a configuration that reduces browser-side security constraints for rendered content.
  • The HTML-REPORT.md file specifies the initialization of the Mermaid engine with securityLevel: "loose". This setting allows more flexible diagram rendering but also permits the execution of scripts within diagrams, which could be exploited if the codebase contains malicious diagram definitions.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 06:50 PM