Integrations

MCP Server

The network itself speaks MCP. Add it as a server, and your agent gains a meta-tool: the ability to evaluate every other tool before using it.

Available tools

Field	Description
search_capabilities	Find ranked capabilities for a task description.
get_capability	Fetch a capability card: interfaces, actions, latest readiness, evidence.
preflight_capability	Check a specific action is ready right now.
report_capability_outcome	Report whether the result was usable after invocation.
explain_readiness_score	Get the human-readable evidence trail behind a score.

Client configuration

Register the server in your MCP client configuration (example below for a stdio transport). Claude Code, Cursor, Cline, and CodeBuddy all follow this pattern.

mcp.json

{
  "mcpServers": {
    "capability-readiness": {
      "command": "python",
      "args": ["-m", "app.mcp_server.server"],
      "env": {
        "PUBLIC_API_BASE_URL": "https://api.example.com",
        "API_KEY": "<your key>"
      }
    }
  }
}

Using the tools

example

# Inside an MCP-enabled agent session:
# the agent calls the tool like any other —

search_capabilities(
  query="extract tables from a PDF into CSV",
  agent_runtime="claude-code",
  risk_tolerance="read_only"
)

# → returns recommended / fallbacks / blocked with scores,
#   evidence levels, and reason codes

A typical agent policy: search once per task, preflight before each call, report outcome after each call. The outcome report is what earns higher-quality recommendations over time.