postrv

postrv / forgemax

Public

Code Mode inspired local sandboxed MCP Gateway - collapses N servers x M tools into 2 tools (~1,000 tokens)

ai-agents ai-tools gateway mcp mcp-gateway
124
6
100% credibility
Found Feb 24, 2026 at 20 stars 6x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

Forgemax is a gateway that condenses access to numerous AI tools from various services into just two efficient functions: search for discovery and execute for safe, chained operations.

How It Works

1
🔍 Discover Forgemax

You hear about a smart helper that lets your AI connect to many services like GitHub or Notion using just two simple commands, saving tons of space in chats.

2
📥 Get it set up

Choose an easy way to add it to your computer, like a quick download that places it ready to use.

3
🔗 Link your services

Copy a sample guide and add your personal logins for the services you want your AI to reach, like code analysis or design tools.

4
🚀 Start the magic gateway

Run it once and it quietly connects everything, ready for your AI to explore.

5
🤖 Hook up your AI chat

Tell your AI helper like Claude or your code editor to use this new gateway, and it starts seeing all the tools.

6
💡 AI discovers and acts

Your AI searches for the right tools, writes quick code to chain them together, and gets results fast without overload.

🎉 Supercharged AI wins

Now your AI handles complex tasks across services seamlessly, making you way more productive.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 20 to 124 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is forgemax?

Forgemax is a Rust-built MCP gateway that aggregates tools from multiple servers—like GitHub, Cloudflare, and Narsil—into just two: `search` for querying a capability manifest and `execute` for running JavaScript against them in a V8 sandbox. It slashes LLM context from 15,000 tokens of schemas to ~1,000 by letting agents write code instead of picking from long lists, inspired by Cloudflare's code mode. Install via npm, Homebrew, or Cargo, configure with a TOML file using env vars for tokens, and run over stdio for code github ai or copilot-like workflows.

Why is it gaining traction?

Agents chain calls in one execution without round-trips, and LLMs excel at code like `narsil.symbols.find({pattern: "handle_*"})` over JSON menus. Security shines with timeouts, heap limits, process isolation, and audit logs—no creds leak to JS. Pre-configs for 11 servers (GitHub repo ops, Playwright, Stripe) make it dead simple versus wiring each MCP endpoint manually.

Who should use this?

Backend devs building AI agents for code github cli, comments, or markdown analysis in VS Code/Cursor/Claude Desktop. Teams modernizing codebases with ollama-local models or integrating github copilot alternatives without token overload. Anyone tired of per-tool schemas in multi-server setups like Narsil + GitHub.

Verdict

Try it for MCP-heavy agents—222 tests, solid README, and easy installs signal quality despite 17 stars and 1.0% credibility. Still early (v0.1.1, FSL license), so test in non-prod first.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.