Rules MCP server

Read from the package source without running it. The installed server may list more.

• enrich_prompt

  Turn a short feature request into a full governed instruction set using project YAML rules, cursor skills (MD/MDC), and knowledge. Includes skillsDisclosure — the agent must open the user-visible reply with that section (skills chosen + why).

• explain_violation

  Explain a rule/violation by ID: rule text, reason, good/bad examples, exceptions, and related knowledge. Use when the user asks why a rule failed or what TS001/SEC001 means.

• generate_artifacts

  Generate agent instruction files from the YAML rule source of truth: .cursor/rules, AGENTS.md, CLAUDE.md, STYLEGUIDE.md, SECURITY.md, CONTRIBUTING.md, PR checklist, and prompt templates. Use when the user asks to sync or regenerate governance artifacts.

• get_context

  Gather relevant YAML rules, Markdown/MDC skills (from cursor skills/), and knowledge for a coding task. Returns skillsDisclosure — paste that block at the start of your reply before answering. Prefer calling this or select_skills first on every coding prompt.

• get_knowledge

  Retrieve project knowledge (architecture, FAQs, module docs). Use when the user asks how the project is structured or about architecture/domain knowledge.

• get_rule

  Fetch full details for one YAML governance rule by ID (e.g. TS001, SEC001). For Markdown skills, use get_skill with a skill slug (e.g. typescript-expert).

• get_skill

  Fetch one skill by slug/id from cursor skills/ (e.g. typescript-expert, nestjs-best-practices/security-validate-all-input). Skills are MD/MDC, not YAML.

• list_rules

  List project coding/governance rules from rules/*.yaml (enforceable standards and policies). For Cursor agent skills (Markdown/MDC under cursor skills/), use list_skills instead.

• list_skills

  List Cursor agent skills loaded from cursor skills/ as Markdown (.md) or MDC (.mdc). Use when the user asks what skills exist or which skill guidance applies.

• ping

  Health check for Rules MCP. Returns server version, phase, and configured paths.

• review_diff

  Review a unified git diff/patch against project rules. Reports violations on new/changed lines. Use when the user asks to review a PR diff, patch, or staged changes for standards compliance.

• score_output

  Produce a governance scorecard (aiScore, category scores, blocking, failedCritical) for code. Use when the user asks for a quality/compliance score.

• search_knowledge

  Ranked search across project knowledge docs (architecture, FAQ, modules). Use when the user asks to find docs about a topic with relevance ranking.

• search_rules

  Search YAML governance rules by keyword (security, logging, python, secrets, etc.). For Cursor skills under cursor skills/, use search_skills.

• search_skills

  Search Markdown/MDC skills under cursor skills/ by keyword (nestjs, typescript, prompt, security, etc.).

• select_skills

  PRIMARY first step for any coding/governance prompt: automatically pick relevant cursor skills for the user's task and return a ready-to-paste 'Skills used' disclosure (id, why, score). Call this (or get_context) before answering, then paste skillsDisclosure at the start of your reply.

• server_info

  Describe Rules MCP capabilities available in the current phase.

• validate_code

  Check a single code snippet against enforceable YAML rules under rules/ and return line-level issues (rule id, message, suggestion). Markdown skills are advisory via list_skills/get_context — they are not machine-validated here. For folders or multiple files, use the multi-file evaluation tool instead.

• validate_paths

  PRIMARY tool when the user wants to evaluate, review, scan, audit, or check source files/folders for enforceable YAML rule violations (any languages: js, ts, py, java, go, …). Reads files from disk, applies matching rules under rules/ as ground truth, and returns every issue with file, line, ruleId, rule text, and suggestion. Skills under cursor skills/ are MD/MDC guidance (use list_skills / get_c

• Code scan71 source files scanned25/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitylocal package, no credentials required12/15
• Maintenancerepository not readable: repo not found3/15
• Maintainer identityno repository or website to verify2/10

Rules MCP: common questions

Is Rules MCP server safe?

With care: it is graded C, so read the findings first (65/100). Read the Rules MCP safety report

How do I install Rules MCP?

It runs on your machine. Copy the Claude Code, Claude Desktop or Cursor config from the install section.

Does Rules MCP need an API key?

No secret keys are declared. It reads 6 settings from the environment.

Is Rules MCP maintained?

The latest release is v0.3.1.