Skip to main content
Glama
claude-rules.md6.54 kB
# GuardianMCP - Comprehensive Rules Template This file contains comprehensive auto-trigger rules for GuardianMCP. Copy the sections you need to your project's `.claude/rules.md` or global `~/.claude/rules.md`. ## Basic Auto-Scan Configuration When working in this project, automatically check for CRITICAL and HIGH severity vulnerabilities in the following scenarios: 1. **Project Start**: When first opening this project, scan dependencies 2. **Security Keywords**: When I mention: security, vulnerability, CVE, audit, exploit, patch 3. **After Package Changes**: After running npm install, npm update, composer install, or composer update 4. **Before Commits**: Before creating git commits, verify no CRITICAL vulnerabilities exist Use the check_vulnerabilities tool with scan_mode="critical-high-only" for automatic scans. Only report CRITICAL and HIGH severity issues unless I explicitly ask for full details. --- ## Detailed Rules (Pick and Choose) ### 1. Project Initialization Auto-Scan When starting work in a project directory that contains package.json or composer.json, automatically run the check_vulnerabilities tool with scan_mode="critical-high-only". Only report findings if CRITICAL or HIGH severity vulnerabilities are detected. Stay silent if no serious vulnerabilities are found. ### 2. Multi-Language Security Keyword Detection When the user mentions any security-related keywords in any language, automatically check for vulnerabilities using scan_mode="critical-high-only". Supported keywords: - English: security, vulnerability, vulnerabilities, CVE, exploit, patch, audit, malware - Latvian: drošība, ievainojamība, drošības risks, ievainojamības - French: sécurité, vulnérabilité, vulnérabilités - Spanish: seguridad, vulnerabilidad, vulnerabilidades - German: Sicherheit, Schwachstelle, Schwachstellen - Russian: безопасность, уязвимость, уязвимости - Italian: sicurezza, vulnerabilità - Portuguese: segurança, vulnerabilidade Report findings clearly with severity levels highlighted. ### 3. Post-Installation Dependency Scanning After successfully running any of these package management commands: - npm install / npm i - npm update - npm ci - yarn install / yarn - yarn upgrade - pnpm install - pnpm update - composer install - composer update - composer require Automatically scan the dependencies for vulnerabilities using scan_mode="critical-high-only". If CRITICAL or HIGH vulnerabilities are found, notify the user with: - Number of vulnerabilities - Package names affected - Recommended immediate actions ### 4. Pre-Commit Security Gate Before creating a git commit (when user asks to commit changes, create a commit, or run git commit), automatically check for CRITICAL severity vulnerabilities. If CRITICAL vulnerabilities are found: 1. Display a clear warning with ⚠️ emoji 2. List the critical vulnerabilities 3. Ask the user: "Critical security vulnerabilities detected. Do you want to proceed with the commit anyway?" 4. Wait for user confirmation before proceeding Use scan_mode="critical-high-only" for this check to keep it fast. ### 5. Build-Time Security Check Before running build commands: - npm run build - npm run production - yarn build - composer build - make build Check for CRITICAL vulnerabilities and warn the user if any are found. Don't block the build, but make the warning very visible. Use scan_mode="critical-high-only". ### 6. Periodic Security Reminders At the start of each new day (first message of the day in a project), if more than 7 days have passed since the last dependency update, remind the user to run a security scan: "It's been a while since the last dependency check. Would you like me to scan for vulnerabilities?" If user agrees, run check_vulnerabilities with scan_mode="critical-high-only". ### 7. When User Adds New Dependencies If you observe the user manually adding a new package to package.json or composer.json (by editing the file directly), suggest running a vulnerability check after they save. For example: "I noticed you added a new dependency. Would you like me to check it for known vulnerabilities?" ### 8. Full Audit on Request When the user explicitly asks for a "full security audit", "complete vulnerability scan", or "detailed security report", use scan_mode="full" to show ALL vulnerabilities including MODERATE and LOW severity issues. --- ## Scan Mode Selection Guide **Use scan_mode="critical-high-only" for:** - Auto-scans (project start, post-install, pre-commit) - Background monitoring - Routine checks - When you want to minimize noise **Use scan_mode="summary" for:** - Quick status checks - Dashboard views - When user asks "how many vulnerabilities" - CI/CD pipelines **Use scan_mode="full" for:** - Comprehensive security audits - When user explicitly asks for "all vulnerabilities" - Before major releases - Compliance reviews --- ## Customization Tips 1. **Adjust aggressiveness**: If auto-scans are too noisy, only trigger on CRITICAL (not HIGH) 2. **Project-specific**: For high-security projects, use scan_mode="full" on all auto-scans 3. **CI/CD integration**: Use scan_mode="summary" for pipeline status checks 4. **Time-based**: Only auto-scan during business hours to avoid notification fatigue 5. **Selective scanning**: Only scan package.json OR composer.json if your project uses one --- ## Example Usage Once these rules are active, the following interactions will trigger automatic scans: **User:** "Let me commit these changes" **Claude:** *Automatically runs vulnerability check before commit* **User:** "npm install express" **Claude:** *Waits for command completion, then scans for vulnerabilities* **User:** "Is our app secure?" **Claude:** *Detects "secure" keyword, runs vulnerability scan* **User:** "Check for security issues" **Claude:** *Runs full vulnerability scan* --- ## Testing Your Rules To verify rules are working: 1. Mention the word "security" in a message 2. Run `npm install` or `composer install` 3. Try to create a git commit 4. Ask "scan my project for vulnerabilities" Claude should automatically use the check_vulnerabilities tool in these scenarios. --- ## Notes - These rules work with both project-level (`.claude/rules.md`) and global (`~/.claude/rules.md`) rules files - More specific project rules take precedence over global rules - You can combine multiple rules - Claude will follow all applicable ones - Adjust language keywords based on your team's primary languages

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Kalvisan/guardian-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server