api-to-mcp-skills

<!-- MCP.com.ai — GitHub Profile README Source reference: README-mcp-ai.md (MCP.com.ai parent project) --> <p align="center"> <img src="https://img.shields.io/badge/MCP-Model%20Context%20Protocol-1193b0?style=for-the-badge" /> <img src="https://img.shields.io/badge/API--First%20for%20AI-OpenAPI%20%E2%86%92%20MCP-da7756?style=for-the-badge" /> </p> <h1 align="center">API-to-MCP (Swagger-to-MCP) Skill</h1> <p align="center"> <b>API-First for AI.</b><br/> Turn existing <b>OpenAPI v3 specs</b> into <b>MCP servers</b> you can deploy, secure, and operate like real software — with HAPI CLI and zero custom code. </p> <p align="center"> <a href="https://mcp.com.ai"><b>Website</b></a> · <a href="https://docs.mcp.com.ai"><b>Docs</b></a> · <a href="https://registry.modelcontextprotocol.io/?q=ai.com.mcp"><b>MCP Servers</b></a> · <a href="https://run.mcp.com.ai"><b>Run MCP</b></a> · <a href="https://hapi.mcp.com.ai"><b>CLI Tool</b></a> </p>

---

🪝 Why this exists

Most orgs don’t have an “AI problem”.
They have an integration and governance problem.

• Your tools live behind APIs
• Your teams ship OpenAPI specs (or should)
• Your agents/IDEs need safe, auditable access
• You need control: identity, policy, rate limits, logs, approvals, isolation

This skill is a focused bridge for OpenAPI → MCP workflows:

• OpenAPI v3 → MCP (standard-driven)
• Deploy anywhere (local, Docker, Cloudflare Workers)
• Security-first patterns (authority outside the model)

---

⚡ What you’ll find here

✅ Skill-aligned workflow

• Environment setup (HAPI CLI + OpenAPI v3 validation)
• Local serve and MCP endpoint verification
• Deployment options (Cloudflare Workers or Docker)
• Verification checklist for operational readiness

✅ Operational patterns

• Safe defaults (timeouts, retries)
• Predictable auth strategies
• Minimal schemas to reduce token bloat

✅ References

• CLI commands, deployment patterns, troubleshooting

---

🚀 Quickstart (Skill Summary)

Requirements

• HAPI CLI (hapi)
• curl
• Optional: docker for container deployment
• Optional: wrangler for Cloudflare Workers

Option A — Start a local MCP server

hapi serve <project-name> \
  --openapi <path-or-url-to-openapi-spec> \
  --url <backend-api-url> \
  --headless \
  --port 3030

Verify health and MCP transport:

curl -s http://localhost:3030/health
curl -s http://localhost:3030/mcp/ping

Option B — Deploy

Cloudflare Workers

hapi login
hapi deploy \
  --openapi <path-or-url-to-openapi-spec> \
  --url <backend-api-url> \
  --name <worker-name> \
  --project <project-name>

Docker (self-hosted)

docker run --name hapi-<project> -d --rm \
  -p 3030:3030 \
  hapimcp/hapi-cli:latest serve <project-name> \
  --openapi <openapi-url> \
  --url <backend-api-url> \
  --port 3030 \
  --headless

---

🔐 Security stance (simple but strict)

Reasoning stays with the model. Authority stays with the system.

We optimize for:

• no long-lived secrets in the model
• scoped, short-lived credentials
• policy enforcement at the boundary
• audit logs by default
• transport that fits the environment (cloud, enterprise, airgap)

---

🧪 Testing & eval (MCP quality is measurable)

Good MCP isn’t just “it runs”. It’s:

• correct tool selection
• stable schemas
• predictable pagination
• safe error behavior
• regression coverage with real prompts

We support evaluation patterns inspired by modern “skills” tooling:

• scenario suites
• tool-call assertions
• latency + cost budgets
• golden traces

---

🧭 Who this is for

• Platform teams building internal AI enablement
• DevOps/SRE who will be on-call for “AI integrations”
• API teams who already own OpenAPI contracts
• Security teams who need guardrails, identity, and auditability
• Enterprise architects tired of one-off agent glue

---

🤝 Contribute

We love contributions that make MCP more operational:

• new MCP server examples
• OpenAPI → MCP mapping improvements
• eval suites for servers
• security patterns (auth, policy, isolation)
• docs that help non-developers succeed

How to contribute

1. Open an issue describing the server or improvement
2. Provide a minimal spec + example prompts
3. Submit PR with tests (when possible)

---

📚 References

• HAPI CLI commands
• Deployment patterns
• Troubleshooting

📣 Stay close

If you want MCP to be more than a demo, follow the ecosystem:

• MCP.com.ai updates (docs + servers)
• HAPI MCP Stack releases (deploy, registry, lifecycle)

Building MCP isn’t the hard part.
Operating MCP safely is the product.

---

<p align="center"> <img src="https://img.shields.io/badge/La%20Rebelion%20Labs-Building%20API--First%20for%20AI-0d1117?style=for-the-badge&labelColor=1193b0&color=da7756" /> </p>