CLI

Every Elliot operation has a CLI command. The same commands are exposed to agents as MCP tools — anything you can run, an agent can run.

elliot init

Scaffold a new connector from a template.

elliot init --template rest-api-key my-api.connector.json

Templates:

• rest-api-key — REST API with API key header auth
• postgres-readonly — PostgreSQL read-only connector
• paginated-rest — REST API with cursor/offset pagination
• openapi-petstore — full Petstore example (docs/tutorials)

Run elliot init --list to print this list.

elliot lint

Validate a connector against the schema and the five principles.

elliot lint my-api.connector.json

Exit code is non-zero on any failure. CI-friendly.

elliot eval

Run an eval suite and report pass/fail with row counts and token estimates.

Eval is deterministic — there is no LLM involved. Each case names a tool and the arguments to call it with; Elliot executes that tool directly against the connector and asserts on the result (row counts, fields present, token size, error codes). This makes eval fast, free, and reproducible in CI.

elliot eval my-api.eval.yaml

An eval suite has a name, the connector it targets, and a list of cases. Each case has an id, a tool_id, the arguments to call it with, and an expect block of assertions:

name: My SaaS Analytics
connector: my-saas
version: "1.0.0"
cases:
  - id: list-customers-by-plan
    description: Filters to only pro-plan customers
    tool_id: list_customers
    arguments: { plan: pro }
    expect:
      no_error: true
      min_rows: 1
      fields_present: [id, name, email, plan]
      max_token_estimate: 800
      all_rows_match: { field: plan, value: pro }

Supported expect keys: no_error (default true), min_rows, max_rows, fields_present, all_rows_match ({ field, value }), error_code, and max_token_estimate. See connectors/my-saas.eval.yaml in the repo for a complete working suite.

elliot connect

Auto-register Elliot with every coding agent on the host.

elliot connect

Detects Claude Code, Cursor, OpenClaw, and Codex and writes the right MCP config for each. Re-run any time you install a new agent.

elliot status

Show the health of plugin, runtime, and Studio.

elliot status

elliot-mcp-plugin       :3000   ✓ healthy
elliot-connector-runtime :3001  ✓ healthy
elliot-studio            :5173  ✓ healthy

elliot deploy

Push a connector into the running runtime (or a hosted runtime).

elliot deploy my-api.connector.json

Same commands, agentic edition

Each of the above is registered with the MCP server as a tool with a verb-first description. An agent inside Claude Code can run lint-connector and run-eval against a connector it just wrote — without leaving the chat.