Heku - Single Server for Any API

One MCP server for any API—no new deployments needed

Unclaimed

Added Jul 2026

rapidthoughtlabs.com

Mcp DeveloperFreemcp-developer

Visit Heku

Share: X in LinkedIn

Add your screenshot here

Image or video shown in this spot

Get Verified · $10 lifetime

What is Heku?

Heku is a dynamic MCP server that eliminates the need to create separate MCP servers for each API integration. Instead of deploying individual HTTP wrapper servers for GitHub, Stripe, Linear, or custom REST APIs, you describe integrations as JSON configs—heku loads them all in a single process. It supports GraphQL, gRPC, and CLI commands, with lazy discovery that keeps cold-start manifests small and token usage low. Built for AI agents and teams managing dozens of internal or third-party APIs without operational overhead.

Explore Heku

Need help implementing Heku - Single Server for Any API?

Find verified specialists who work with Heku - Single Server for Any API

Browse specialists

Key Features of Heku

Dynamic MCP server supporting HTTP, GraphQL, and gRPC integrations
JSON-based config for instant API onboarding
Lazy discovery keeps cold-start manifests small
Hot reload for adding tools without restarting
GraphQL introspection for automatic tool discovery
Hosted console for testing tools locally
Environment-based credential management
O(1) cold connect regardless of config count
Single process handles unlimited integrations
Publish configs to heku hub for community sharing

Who Should Use Heku?

Wire multiple APIs (GitHub, Slack, Linear, Stripe) to AI agents without separate servers

Manage hundreds of internal REST endpoints as agent tools

Deploy a hosted gateway for multi-agent access over HTTP

Publish and share API integrations via heku hub

Reduce MCP manifest size and token usage at agent cold-start

Onboard new APIs in seconds with JSON configs instead of new repos

Heku: Pros & Cons

✓Pros

One server replaces ~95% of typical MCP server use cases (HTTP proxies and CLI wrappers)
Dramatic reduction in manifest size (529–757 tokens vs ~30,000 with 10 separate servers)
Lazy discovery ensures O(1) cold-start cost regardless of config count
Hot reload allows adding tools without restarting or redeploying
GraphQL and gRPC introspection automate tool discovery
Credentials stored in .env files, not hardcoded in configs
Simple JSON syntax—no coding required for most integrations
Hosted console for quick testing and development

Frequently Asked Questions about Heku

What is heku and why do I need it?

Heku is a dynamic MCP server that consolidates multiple API integrations into a single process. Instead of deploying separate MCP servers for each API (GitHub, Stripe, Linear, etc.), you describe each integration as JSON and heku serves them all as tools your agent can call. This eliminates operational overhead and reduces manifest bloat.

How does lazy discovery work and what does it save?

Heku uses lazy discovery to keep the cold-start manifest small. Instead of loading all tool schemas at connect time, it only loads what the agent needs. This reduces token usage from ~30,000 tokens (with 10 separate servers) to 529–757 tokens, and the cost stays O(1) even with hundreds of configs.

What types of APIs can heku integrate with?

Heku supports HTTP/REST APIs, GraphQL endpoints, gRPC services, and CLI commands. For GraphQL and gRPC, heku can introspect the source to auto-discover tools, so you don't need to manually define them in your config.

How do I get started with heku?

Install with `npm install -g @rapidthoughtlabs/heku`, write a JSON config describing your API connector type and endpoint, then run `heku start`. You can test tools immediately with the hosted console at localhost. New APIs can be added by dropping a new config file—no redeploy required.

Can I share my heku integrations with others?

Yes, you can publish configs to heku hub and others can install them with one command (e.g., `heku install @rtl/github`). This lets teams reuse integrations without duplicating configs or servers.