MCP Integration

Connect your agents to organizational intelligence.

OTP exposes all OOS data via the Model Context Protocol. Any AI agent, Claude, GPT, or custom, can query organizational intelligence natively.

Not just a website. A protocol.

Quick start → View tools

The protocol

Every data source should be an MCP server. Including your operating system.

The Model Context Protocol lets AI agents access external data sources as native tools. Instead of scraping websites or parsing APIs, agents call MCP tools directly, with structured inputs and structured outputs.

OTP's MCP server turns the entire Intelligence Graph into a tool your agents can use. Browse published OOS files. Search across claims. Compare coordination patterns between organizations. All from inside your agent's workflow.

Your agent doesn't need to leave its environment to access organizational intelligence. It's already there.

Available tools

What your agent can do.

Each tool is a structured operation your agent can call directly via MCP.

publish_oos()

Publish an OOS file programmatically. Upload structured claims, set visibility, and join the Intelligence Graph.

browse_oos()

Browse published OOS files with filters for industry, size, template, and quality tier. Paginated results with metadata.

get_oos()

Retrieve a specific OOS by ID. Returns full claim structure, confidence distributions, and organizational metadata.

compare_oos()

Compare two OOS files side by side. Returns shared claims, unique claims, similarity scores, and coordination gaps.

search_claims()

Search across all published claims using natural language. Filter by confidence, evidence level, section, and industry.

get_publishers()

Browse organizations on the platform. Filter by industry, quality tier, and agentic maturity level.

Quick start

Add OTP to your agent in 30 seconds.

Recommended: Remote connection (no install)

Hosted · paid

On a paid plan you can connect with one hosted URL — no npx, no terminal. In Claude, go to Settings → Connectors → Add custom connector, paste https://orgtp.com/api/mcp, and sign in — one click, no token to copy. Prefer a token URL (for clients without sign-in support)? Generate one from Settings → API. The steps below are the self-hosted alternative.

Self-hosted: add the MCP server to your config

{
  "mcpServers": {
    "otp": {
      "command": "npx",
      "args": ["-y", "otp-mcp-server"],
      "env": {
        "OTP_API_KEY": "your-api-key"
      }
    }
  }
}

Get your API key

Start querying

Your agent can now access the full Intelligence Graph. Try these natural language prompts:

"Browse all published OOS files in the AI services industry"

"Search for claims about agent escalation patterns"

"Compare Sneeze It's OOS with the Expert Coach template"

Use cases

What agents do with organizational intelligence.

Agent Onboarding

Agent reads the organization’s OOS on first run to understand coordination rules, authority boundaries, and escalation paths. No manual configuration.

Cross-Org Discovery

Agent searches for organizations with similar coordination patterns. Compatibility signals guide vendor selection, partnerships, and inter-org coordination.

Continuous Learning

Agent periodically queries the Intelligence Graph for new patterns in its industry. Adapts coordination protocols as the network’s collective intelligence grows.

Compliance Checking

Agent validates its actions against published authority boundaries. Before taking an action, it checks whether the OOS permits it, programmatically.