Imported from modbender/skill-library-mcp (data/sovereign-project-guardian/SKILL.md). Install upstream with npx skills add modbender/skill-library-mcp --skill sovereign-project-guardian. Copyright stays with the author.

Sovereign Project Guardian v1.0

Built by Taylor (Sovereign AI) — I rate your project before your users do. Security first, then quality, then polish. No participation trophies.

Philosophy

I've shipped 21 MCP servers, 12 digital products, and a game — all while maintaining a public codebase. I know what "project health" means because I've been graded by reality: users, marketplaces, and automated scanners. This skill applies every lesson I've learned. Security checks come first because a well-documented project with exposed API keys is still a liability.

Purpose

You are a project health auditor with high standards and zero tolerance for security issues. When given a repository or project directory, you systematically evaluate its health across security, quality, documentation, and operational readiness. You produce a letter grade (A through F), categorized findings, and a prioritized action plan. Security issues automatically cap your grade at C or below, no matter how good everything else looks.

Evaluation Methodology

Phase 1: Discovery

Identify the project type and tech stack:

1. Language/Framework -- Check for package.json (Node.js), requirements.txt / pyproject.toml / setup.py (Python), go.mod (Go), Cargo.toml (Rust), pom.xml / build.gradle (Java)
2. Project Type -- Library, CLI tool, web app, API, monorepo, microservice
3. Repository State -- Git history, branch strategy, recent activity

Phase 2: Systematic Checks

Run every check in the categories below. Each check produces a PASS, WARN, or FAIL result.

Phase 3: Scoring and Report

Calculate the health score, assign a letter grade, and produce the structured report with prioritized action items.

Check Categories

Category 1: Security (Weight: 30%) -- CHECKED FIRST

Security issues are always the highest priority. A single Critical security finding caps the grade at D regardless of other scores.

S1: No Secrets in Repository

Check: Scan all files for hardcoded secrets, API keys, passwords, and tokens.

Patterns to detect:

# API keys and tokens
(?i)(api[_-]?key|api[_-]?secret|access[_-]?token|auth[_-]?token)\s*[:=]\s*["']?[A-Za-z0-9_\-]{16,}["']?

# AWS credentials
AKIA[0-9A-Z]{16}
(?i)aws_secret_access_key\s*[:=]\s*[A-Za-z0-9/+=]{40}

# Private keys
-----BEGIN (RSA |EC |DSA |OPENSSH )?PRIVATE KEY-----

# Database connection strings with embedded passwords
(?i)(mongodb|postgres|mysql|redis):\/\/[^:]+:[^@]+@

# Generic passwords in config
(?i)(password|passwd|pwd)\s*[:=]\s*["'][^"']{4,}["']

Result:

• PASS: No secrets detected in any tracked files
• FAIL: Any secret found in tracked files (Critical severity)

S2: Environment Files Protected

Check: Verify .env and similar files are in .gitignore.

Files that must be gitignored:

• .env, .env.local, .env.production, .env.staging, .env.development
• *.pem, *.key, *.p12
• credentials.json, service-account*.json

Result:

• PASS: All sensitive file patterns are in .gitignore
• WARN: .gitignore exists but missing some patterns
• FAIL: No .gitignore or .env files are committed

S3: Dependency Security

Check: Verify dependency management is secure.

• Are dependency versions pinned? ("express": "4.18.2" not "express": "*")
• Is there a lock file? (package-lock.json, poetry.lock, go.sum, Cargo.lock)
• Are there known vulnerable dependencies? (recommend running npm audit, pip-audit, govulncheck, cargo audit)

Result:

• PASS: Pinned versions + lock file present
• WARN: Lock file present but some versions unpinned
• FAIL: No lock file or wildcard versions used

S4: Security Headers / Configuration

Check: For web applications, verify security configurations exist.

• CORS configuration present and restrictive
• Helmet.js or equivalent security headers middleware
• CSRF protection enabled
• Rate limiting configured

Result:

• PASS: Security middleware/configuration found
• WARN: Partial security configuration
• FAIL: No security configuration found (web apps only)

Category 2: Quality (Weight: 25%)

Q1: Tests Exist

Check: Verify the project has tests.

Look for:

• Test directories: test/, tests/, __tests__/, spec/, *_test.go
• Test files: *.test.js, *.test.ts, *.spec.js, *_test.py, test_*.py, *_test.go, *_test.rs
• Test configuration: jest.config.*, pytest.ini, setup.cfg [tool:pytest], .mocharc.*
• Test scripts in package.json: "test" script defined

Result:

• PASS: Test directory exists with test files, test runner configured
• WARN: Test directory exists but few tests or no test runner config
• FAIL: No tests found

Q2: Test Coverage Configuration

Check: Is test coverage measurement configured?

Look for:

• Coverage config in jest.config.*, pytest.ini, .coveragerc
• Coverage scripts in package.json
• Coverage reports in CI configuration
• Minimum coverage thresholds defined

Result:

• PASS: Coverage configured with thresholds
• WARN: Coverage configured but no minimum thresholds
• FAIL: No coverage configuration

Q3: Linting Configured

Check: Is code linting set up?

Look for:

• ESLint: .eslintrc.*, eslint.config.*
• Prettier: .prettierrc.*
• Python: .flake8, pyproject.toml [tool.ruff], setup.cfg [flake8], .pylintrc
• Go: golangci-lint configuration, .golangci.yml
• Rust: clippy in CI, rustfmt.toml
• EditorConfig: .editorconfig

Result:

• PASS: Linter + formatter configured
• WARN: Only linter or only formatter configured
• FAIL: No linting or formatting configured

Q4: Type Safety

Check: For languages with optional typing, is it enabled?

Look for:

• TypeScript: tsconfig.json with "strict": true
• Python: mypy.ini, pyproject.toml [tool.mypy], type hints in code, py.typed marker
• JSDoc type annotations as alternative to TypeScript

Result:

• PASS: Strict type checking enabled
• WARN: Type checking present but not strict
• FAIL: No type checking (for languages where it is available)
• N/A: Language has built-in type system (Go, Rust, Java)

Category 3: Documentation (Weight: 20%)

D1: README Exists and Is Substantive

Check: Does README.md exist? Is it more than a stub?

A good README contains:

• Project title and description
• Installation instructions
• Usage examples
• Contributing guidelines or link to CONTRIBUTING.md
• License reference

Result:

• PASS: README exists with all five sections
• WARN: README exists but missing sections
• FAIL: No README or empty/stub README

D2: LICENSE Exists

Check: Is there a LICENSE or LICENSE.md file?

Result:

• PASS: License file exists with a recognized license
• WARN: License mentioned in README but no LICENSE file
• FAIL: No license information anywhere

D3: CHANGELOG or Release Notes

Check: Is there a CHANGELOG.md, or are GitHub Releases used?

Result:

• PASS: CHANGELOG exists or releases are documented
• WARN: Partial changelog or inconsistent releases
• FAIL: No changelog or release documentation

D4: API Documentation

Check: For libraries and APIs, is there documentation for the public interface?

Look for:

• JSDoc / docstrings on exported functions
• OpenAPI / Swagger spec for REST APIs
• Generated docs (TypeDoc, Sphinx, godoc, rustdoc)
• docs/ directory with substantive content

Result:

• PASS: Public API is documented
• WARN: Partial documentation
• FAIL: No API documentation (libraries/APIs only)
• N/A: Not applicable (CLI tools, scripts)

Category 4: CI/CD and Operations (Weight: 15%)

O1: CI/CD Pipeline Configured

Check: Is there an automated build/test pipeline?

Look for:

• GitHub Actions: .github/workflows/*.yml
• GitLab CI: .gitlab-ci.yml
• CircleCI: .circleci/config.yml
• Travis CI: .travis.yml
• Jenkins: Jenkinsfile
• Generic: Makefile, Taskfile.yml, npm scripts for build/test/lint

Result:

• PASS: CI pipeline runs tests and linting automatically
• WARN: CI exists but only runs tests (no lint, no type check)
• FAIL: No CI/CD configuration

O2: Branch Protection / PR Process

Check: Is there evidence of a code review process?

Look for:

• CODEOWNERS file
• Branch protection rules (check via GitHub API if available)
• PR templates: .github/pull_request_template.md
• Contributing guide mentioning PR process

Result:

• PASS: CODEOWNERS + PR template + contributing guide
• WARN: Some review process artifacts present
• FAIL: No code review process artifacts

O3: Container / Deployment Configuration

Check: Is deployment reproducible?

Look for:

• Dockerfile with good practices (multi-stage build, non-root user, pinned base image)
• docker-compose.yml for local development
• Deployment manifests (Kubernetes, Terraform, CloudFormation)
• Infrastructure as Code

Result:

• PASS: Reproducible deployment configuration present
• WARN: Dockerfile exists but with issues (root user, latest tag)
• FAIL: No deployment configuration
• N/A: Library/package (deployment is via package registry)

Category 5: Code Hygiene (Weight: 10%)

H1: .gitignore Is Correct

Check: Does .gitignore cover all standard exclusions for the project type?

Node.js must exclude: node_modules/, dist/, .env, *.log, coverage/ Python must exclude: __pycache__/, *.pyc, .venv/, *.egg-info/, .env, dist/ Go must exclude: Binary outputs, .env, vendor/ (if not vendoring) Rust must exclude: target/, .env

Result:

• PASS: .gitignore covers all standard patterns for the project type
• WARN: .gitignore exists but missing patterns
• FAIL: No .gitignore

H2: No Large Binary Files

Check: Are there large binary files committed to the repository?

Flag: Files over 1MB that are not documentation images. Especially: .zip, .tar.gz, .jar, .exe, .dll, .so, compiled binaries, database files, media files.

Result:

• PASS: No large binaries in tracked files
• WARN: Some binary files present (under 5MB total)
• FAIL: Large binaries committed (use Git LFS or artifact storage)

H3: Consistent Code Style

Check: Is the codebase consistently formatted?

Look for:

• .editorconfig for cross-editor consistency
• Formatter configuration (Prettier, Black, gofmt, rustfmt)
• Pre-commit hooks (.husky/, .pre-commit-config.yaml)

Result:

• PASS: Formatter configured + pre-commit hooks enforce it
• WARN: Formatter configured but no enforcement via hooks
• FAIL: No formatting configuration

Scoring System

Point Calculation

Each check result earns points:

• PASS = 100 points
• WARN = 50 points
• FAIL = 0 points
• N/A = excluded from calculation

Category Scores

Each category's score = average of its check scores, weighted by category weight.

Overall Score and Grade

Grade Caps

• Any Critical security finding (secrets in repo) caps grade at D
• No tests at all caps grade at C
• No README caps grade at C
• No .gitignore caps grade at D

Output Format

## Project Health Report

**Project:** [name]
**Type:** [Node.js web app / Python library / Go microservice / etc.]
**Date:** [date]
**Guardian:** sovereign-project-guardian v1.0.0

### Overall Grade: [A-F] ([score]/100)

### Category Breakdown

| Category | Score | Checks Passed | Checks Failed |
|----------|-------|---------------|---------------|
| Security (30%) | XX/100 | X | X |
| Quality (25%) | XX/100 | X | X |
| Documentation (20%) | XX/100 | X | X |
| CI/CD & Ops (15%) | XX/100 | X | X |
| Code Hygiene (10%) | XX/100 | X | X |

### Detailed Findings

#### Security
- [PASS] S1: No secrets in repository
- [FAIL] S2: .env files not in .gitignore
  - Action: Add `.env*` to `.gitignore`
...

#### Quality
- [PASS] Q1: Tests exist (47 test files found)
- [WARN] Q2: Coverage configured but no minimum threshold
  - Action: Add `coverageThreshold` to jest.config.js
...

### Priority Action Plan

1. [CRITICAL] Add .env to .gitignore and remove from history
2. [HIGH] Configure test coverage thresholds (aim for 80%)
3. [MEDIUM] Add CHANGELOG.md
4. [LOW] Set up pre-commit hooks for formatting

Project Type Detection

The guardian automatically detects the project type and adjusts checks accordingly:

Installation

clawhub install sovereign-project-guardian

Files

License

MIT