lazarus-mcp
Enables automatic diagnosis, safe repair, and evidence generation for broken Node and Python repositories, providing scan, autopsy, resurrect, and evidence-pack tools via MCP.
README
Lazarus MCP
Lazarus MCP autopsies broken Node/Python repos, applies safe resurrection playbooks, reruns the pipeline, and writes judge-ready proof.
Why It Exists
Hackathon judges and maintainers need more than an AI claim. They need evidence: what failed, why it failed, what changed, and whether build/test is green afterward. Lazarus keeps the scope narrow and auditable: local Node and Python repositories only, conservative patches only, no remote pushes.
Quickstart
npm install
npm run build
npm test
Try a local fixture path:
npm run dev -- scan fixtures/dead-node-missing-build
npm run dev -- autopsy fixtures/dead-node-missing-build
npm run dev -- resurrect fixtures/dead-node-missing-build --safe
npm run dev -- evidence-pack fixtures/dead-node-missing-build
Try a GitHub HTTPS URL:
npm run dev -- scan https://github.com/<owner>/<repo>
npm run dev -- autopsy https://github.com/<owner>/<repo>
npm run dev -- resurrect https://github.com/<owner>/<repo> --safe
npm run dev -- evidence-pack https://github.com/<owner>/<repo>
GitHub URLs are cloned into .lazarus-workspaces/<repo-name>-<timestamp>. Lazarus creates local branches only, never pushes upstream, and never modifies remotes.
CLI Usage
lazarus scan <path-or-github-url>
lazarus autopsy <path-or-github-url>
lazarus resurrect <path-or-github-url> [--safe] [--branch <name>]
lazarus evidence-pack <path-or-github-url>
scan detects stack, package manager, config files, scripts, test runner hints, confidence, and health hints. autopsy runs safe install/build/test commands with timeouts and structured log classification. resurrect applies high-confidence playbooks, reruns the pipeline, and never pushes upstream. evidence-pack writes RESURRECTION_REPORT.md, AI_JUDGES.md, and evidence/summary.json.
MCP Tools
The stdio adapter in src/server.ts exposes:
scan_repoautopsyresurrectevidence_pack
The adapter is intentionally thin so the CLI core stays package-agnostic and green if MCP package APIs shift.
Proof It Works
tests/detect.test.tscovers stack and pytest detection.tests/autopsy.test.tscovers log classification and missing-script autopsy.tests/resurrect.node.test.tsresurrectsfixtures/dead-node-esm-cjsandfixtures/dead-node-missing-build.tests/resurrect.python.test.tsresurrectsfixtures/dead-python-missing-pytest.tests/report.test.tsverifies judge evidence generation.- CI runs
npm install,npm run build, andnpm test.
Demo Flow
- Show a broken fixture repository.
- Run
lazarus autopsy <path>and point to the failing phase and error class. - Run
lazarus resurrect <path> --safe. - Show after metrics are green.
- Run
lazarus evidence-pack <path>and open the generated report files.
Current MVP Scope
- Supported stacks: Node and Python.
- Supported input: local path or GitHub HTTPS URL.
- Supported Node fixes: missing build script, conservative ESM/CJS package type repair, strongly implied TypeScript dependency.
- Supported Python fixes: missing pytest dependency when tests/config imply pytest.
- No frontend, OAuth, remote HTTP server, broad language support, or remote git push.
Safety
- Commands time out after 120 seconds by default.
- Logs are redacted for obvious secrets.
- Lazarus never force-pushes, never pushes, and never modifies remotes.
- Low-confidence fixes are reported as suggestions instead of mutations.
- Applied changes are intentionally small and auditable.
Screenshots / Logs
Terminal demo placeholders live in docs/DEMO_SCRIPT.md and docs/VIDEO_STORYBOARD.md. The machine-readable run summary lives in evidence/summary.json.
Landing Page
The Vercel-ready hackathon landing page lives in apps/web.
npm run landing:dev
npm run landing:build
For Vercel, connect the GitHub repository, set the project root to apps/web, use the Next.js framework preset, and keep the build command as npm run build.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.