Log inSubmit server
Locus logo

Locus

Connects AI agents to pay-per-use web research, data lookups, APIs, and hosted Workflows through OAuth.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Locus

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "locus-3": {
      "command": "npx",
      "args": [
        "-y",
        "plugins"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

We couldn’t confirm it starts correctly

We ran the install command but it didn't respond within our test window — this can mean a slow first-time install rather than a real problem.

npx -y plugins

Process exited before answering initialize.

This is an experimental automated check and can have false negatives — missing environment variables, a slow cold install, etc. It doesn’t necessarily mean something’s wrong. Last checked 3d ago.

Overview

  • web-research
  • data-extraction
  • api
  • workflows

The Locus MCP server connects agents to pay-per-use web research, data lookups, and metered API endpoints through a streamable HTTP MCP connection. Authentication uses browser-based OAuth, while usage is charged to the authenticated workspace's prepaid credits. Agents can retrieve operating guidance, route requests to available providers, and use hosted Workflow capabilities when enabled. Reach for it when an agent needs cited research or paid data access without storing API keys in client configuration.

Use cases

  • Research the web with cited results
  • Look up paid data through metered APIs
  • Route agent requests across provider endpoints
  • Run bounded hosted Workflows
  • Retrieve versioned operating guidance

Key features

  • Streamable HTTP MCP connection
  • Browser OAuth authentication
  • Cited web research
  • Paid data and API lookups
  • Workspace credit billing
  • Hosted Workflow execution

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 3d ago

Recent health check succeeded.

5/5 checks healthy over the last 45d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 16d ago via OSV.dev · plugins (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

npm downloads64kPackage downloads in the last 30 days.
Last commit8d agoMost recent push to the default branch.
Install checkInconclusiveDidn't respond in our test window — often a slow first install.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How Locus works

What Locus MCP server does

The Locus MCP server gives coding and other AI agents access to pay-per-use services through one authenticated MCP connection. Its supported use cases include cited web research, paid data lookups, and requests to metered provider endpoints. Charges are applied to the authenticated workspace's prepaid credit balance rather than to credentials stored in the client.

The repository also supplies three Agent Skills. The main usage skill explains routing, cost estimates, idempotent billing, error handling, and spend controls. The setup skill covers account onboarding. The workflows skill describes hosted Workflow authoring, fixtures, bounded pilots, immutable versions, and recovery of durable runs.

How it works

The Locus MCP server uses streamable HTTP at https://api.paywithlocus.com/api/credits/mcp. A client establishes the connection and completes OAuth in a browser. The repository does not provide credential values or require an API key in configuration. Tool requests carry the arguments supplied by the agent and are billed to the authorized workspace.

An MCP-only client can call get_locus_guide to retrieve the versioned operating and reference material published for the connection. Hosted Workflow tools appear only when the backend capability is available and the current authorization can run them. A Workflow runs as a constrained child of the existing connection rather than as another registered agent.

Setup and configuration

Install the repository's plugin or add the production MCP URL directly in a compatible client. The documented setup paths cover Claude Code, Codex, Cline, Cursor, Claude Cowork, claude.ai, ChatGPT.com, Grok, Hermes, OpenClaw, Muse, and Instinct. Clients that support skills can also install the repository with npx skills add locus-technologies/locus-pro-plugin and select the locus, locus-setup, and locus-workflows skills.

For a plain MCP connection, use the production URL and complete browser OAuth. No clone, build step, environment variable, or API key is required for the documented Cline setup. The plugin-managed configurations may send an X-Source-Name header for anonymous adoption telemetry; the header is optional, and clients that cannot set it can still connect.

Tools and capabilities

The material documents access to:

  • Cited web research.
  • Paid data and API lookups.
  • Metered provider endpoints billed to workspace credits.
  • get_locus_guide for remote operating guidance.
  • Hosted Workflows with approved TypeScript and reviewed catalog bindings, when enabled.
  • Workflow fixtures, bounded pilots, versioning, inspection, cancellation, and resume.

Provider availability, connection scope, enablement, contracts, and run budgets determine which requests can execute. The usage guidance helps agents obtain cost quotes and recover safely from errors before spending credits.

Limitations and notes

The Locus MCP server does not imply access to every provider or endpoint. Hosted Workflows do not automatically include arbitrary npm dependencies, external connector credentials, scheduling, or publication. Tenant-imported and unverified external bindings need exceptional platform approval, and a server deny override can stop a binding.

Human-owned accounts manage tools and credits in the dashboard. Agent-owned accounts through AgentID are available for headless runtimes and are the only setup path described as supporting agent-driven capability selection and a user-requested Stripe funding handoff. The repository's public guide does not include customer-owned Workflow source or invocation guides; those remain private account artifacts.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Locus

Locus is an MCP server that connects AI agents to cited web research, paid data lookups, and metered provider API endpoints. It authenticates through browser OAuth and bills an authorized workspace's prepaid credits. Its documented tools include research and lookup access, `get_locus_guide`, and hosted Workflow capabilities when enabled.

Install the repository plugin or add `https://api.paywithlocus.com/api/credits/mcp` as a streamable HTTP MCP server. Complete browser OAuth, then optionally install the three Agent Skills with `npx skills add locus-technologies/locus-pro-plugin`. The README provides client-specific commands for Claude Code, Codex, Cline, and Cursor.

No, Locus uses browser-based OAuth instead of an API key in client configuration. Sign in through the browser consent flow, and tool calls are billed to the authenticated workspace. The repository contains no credential values.

Locus bills requests to the authenticated workspace's prepaid credit balance. Its usage guidance covers cost quotes, idempotent billing, spend safety, and error recovery. Credits and enabled tools are managed through the Locus dashboard.

Yes, when the backend capability is live and the current connection is authorized to execute it. Workflows support approved TypeScript, reviewed bindings, bounded concurrency, fixtures, pilots, inspection, cancellation, and resume. Arbitrary dependencies, scheduling, external credentials, and publication are not implied.

Related MCP servers

All in Search & Data Extraction All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewLocus AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/locus-3?style=directory)](https://allmcps.com/mcp/locus-3)
HTML Embed
<a href="https://allmcps.com/mcp/locus-3"><img src="https://allmcps.com/api/badge/locus-3?style=directory" alt="Locus on AllMCPs" /></a>