What is Toolport?
Toolport is a free, open source, local MCP (Model Context Protocol) gateway that centralizes how AI agents connect to external tools. Instead of configuring the same MCP servers—connectors that give AI agents access to services like GitHub, Slack, or databases—separately for each client, Toolport lets you set up a server once and share it across Claude, Cursor, VS Code, Codex, and 29 other supported agents. It runs entirely on your machine, requires no account and no cloud, and is available for Windows, macOS, and Linux.
Who it's for
- Individual AI power users who juggle multiple agents and are tired of pasting the same server configs and API keys into each one.
- Token-conscious developers who want to cut context bloat—Toolport replaces hundreds of tool definitions with a handful of meta-tools, reducing token usage by up to 91% at the same task success rate.
- Security-minded teams who need governance over tool access, with per-tool toggles, integrity checks against tampering, and secrets stored in the OS keychain rather than client configs.
Key features
One gateway for every agent
Set up and authenticate an MCP server once in Toolport, and it appears in every supported agent with hot toggles and no restarts. Import servers your agents already have, or add new ones directly—no account, no cloud, no repeated configuration.
Tool-integrity checks
Toolport fingerprints every tool and flags rug pulls (a definition changing after you approved it) and tool poisoning (hidden instructions in a description). These checks are on by default and run entirely locally, so nothing leaves your machine.
Secrets in the keychain
API keys live in your OS keychain and are injected at runtime. They never sit in config files or in the cloud, which means your credentials stay under your control even when you're sharing servers across multiple agents.
Per-tool governance and observability
Toggle any tool on or off individually, or hide every destructive tool from every agent with a single switch. Built-in live observability shows per-server latency, error rates, and a full audit trail of every tool call.
What stands out
Toolport stops the token tax without sacrificing capability—your agent sees a handful of meta-tools, not hundreds of definitions.
That's the core insight: instead of dumping every tool definition into context on every request, Toolport exposes a few meta-tools the agent searches on demand. Benchmarked on a frontier model, this approach delivers up to 91% fewer tokens at the same task success, graded for correct answers. Combined with local-only operation, keychain-based secrets, and tamper detection, it's a rare combination of efficiency, security, and open source transparency (MIT licensed).
Worth checking out if…
You're running multiple AI agents and feel the pain of duplicated configs, ballooning token usage, or scattered API keys. Toolport is especially compelling if you care about security—the integrity checks and keychain storage are meaningful safeguards—or if you're coordinating tools across a small team, since the Teams tier supports up to 5 people free with per-person keys staying on their own machines. If you're on a single agent with a handful of tools, the overhead might not justify itself, but for anyone managing several clients or servers, the setup-once-share-everywhere model is a genuine time and cost saver.
