# Warp MCP connector

Connect Warp to Elaichi and Claude, ChatGPT, Cursor or any MCP client can start and follow Warp agent runs, read transcripts, and manage schedules, with each person acting under their own Warp access and every call logged.

Source: https://elaichi.ai/connectors/warp/

## Facts

| | |
| --- | --- |
| Application | Warp |
| Category | Artificial Intelligence |
| AI tools | 27 |
| Authentication | Connects with an API key |
| Needs your own OAuth app | No |
| MCP endpoint | https://api.elaichi.ai/mcp |
| Works with | Claude, ChatGPT, Cursor, any MCP client, and the Elaichi Agent |
| Tools advertised by name | Yes |

## What you can ask once Warp is connected

- Summarize yesterday's Warp agent runs by agent.
- Which Warp agent schedules created runs this week?
- Show the transcript for our longest Warp run.

## Connect Warp in Elaichi

This happens once for the organization, before any client is involved.

1. Open Connections, choose Add connection, and pick Warp.
2. Optionally set Share with, then press Connect.
3. Paste a Warp API key. One person generates a token in Warp and pastes it once. Everyone else works through Share with, and never sees it.

Credentials are vaulted and nobody, including the AI, reads them back. The connection becomes a toolbox immediately, so you can curate which Warp tools are exposed, rename them, or freeze arguments before anyone points a client at it.

## Connect Warp to Claude

Endpoint: https://api.elaichi.ai/mcp

1. Open Customize, then Connectors.
2. Press Add.
3. Name it, paste the MCP server URL, then Continue.
4. Sign in and approve.

On Team and Enterprise, an Owner adds it once. Everyone else turns it on for themselves.

## Connect Warp to ChatGPT

Endpoint: https://api.elaichi.ai/mcp

1. Open Plugins, then press the + button.
2. Name it and paste the endpoint into Server URL.
3. Leave Authentication on OAuth, then tick the risk acknowledgement.
4. Press Create, then sign in and approve.

Works on the web today. The plugin directory lives at chatgpt.com/plugins.

## Connect Warp to Cursor

Endpoint: https://api.elaichi.ai/mcp

1. Open `~/.cursor/mcp.json`.
2. Add the endpoint under `mcpServers`.
3. Reload Cursor, then sign in and approve.

Set up per machine, so repeat it on each computer you work from.

## Connect Warp to any MCP client

Endpoint: https://api.elaichi.ai/mcp

1. Add the endpoint as a remote MCP server.
2. Sign in and approve.

The Elaichi Agent already has these tools, with nothing to set up.

## What the consent screen decides

Only Read is granted by default, which is not enough to call a Warp tool. Over MCP there is no trusted place to confirm a write in the moment, so the consent screen is the standing approval rather than a formality. Grant Read and Run tools. Think hard before granting Delete, which reaches into connected apps and cannot be undone.

## What teams do with Warp through Elaichi

### Start a Warp agent run from chat

Engineering. Describe the task in Claude, pick the Warp agent, environment and model, and kick off the run without switching windows. Check back later and ask how it is going.

### Catch up on what an agent did

Engineering. Pull the transcript and timeline of a Warp run, ask for a plain summary of the changes, and send a follow-up message to the agent if something needs clarifying.

### Keep scheduled agent runs on track

Platform. Create a Warp agent schedule for recurring maintenance, pause it during a freeze, resume it afterward, and adjust the timing when the team's rhythm changes.

### Spot stuck or failing runs early

Engineering management. List agent runs across the team, find the ones that have been going too long or ended badly, and cancel what is stuck before it burns more compute.

### Reconstruct a run that misbehaved

Support. When a user reports a bad result, pull the run conversation, harness support transcripts and artifacts together so the problem is understood before anyone replies.

### Know which agent identities exist

IT and security. List the agent identities, environments and connected self-hosted workers in Warp, and create a new identity for a project without hunting through settings.

## Frequently asked questions

### How do I connect Warp to Claude?

First connect Warp in Elaichi by pasting a Warp API key when prompted; there is no OAuth application to register and no client ID or secret to generate. Then in Claude open Customize, then Connectors, then Add, and paste https://api.elaichi.ai/mcp as the endpoint. Sign in with your Elaichi account and Warp appears in Claude, ready to use.

### Does Warp work with ChatGPT and Cursor as well as Claude?

Yes. Once Warp is connected in Elaichi, the same endpoint, https://api.elaichi.ai/mcp, works in Claude, ChatGPT, Cursor, any other MCP client, and the Elaichi Agent. You connect Warp once and every client you use picks it up.

### What can an AI agent actually do with my Warp data?

With Warp connected through Elaichi, an AI agent can list your Warp agents, start a run, check its status, cancel it, read the transcript, timeline and conversation, and send a follow-up. It can also create, update, pause, resume or delete agent schedules, look up environments, models, artifacts, agent identities and connected self-hosted workers, and read harness support transcripts. It cannot do anything in Warp beyond those actions.

### Does connecting Warp give the AI access to all my agents and runs?

No. An AI client only reaches what the Warp API key used to connect can already reach, so access follows the person who connected and the permissions their Warp account has. Elaichi can narrow that further with restrictions on individual actions, but it can never widen what Warp itself allows.

### Can my team share one Warp connection?

Yes. One person connects Warp in Elaichi and shares the connection with a team, and nobody else ever handles or sees the Warp API key. Each teammate still signs in to Elaichi as themselves, so the audit log names the actual person behind every Warp run started, schedule changed or transcript read.

### Can I stop an agent from deleting or changing things in Warp?

Yes. Restrictions in Elaichi apply per action, so you can allow reading Warp runs and transcripts while blocking the ability to start runs, cancel them, or delete and edit schedules. A blocked action is never advertised to Claude, ChatGPT or Cursor, so no prompt, however worded, can reach it.

### What happens to a Warp connection when someone leaves?

When you offboard someone in Elaichi, their access to Warp through every AI client ends at once. If they had shared a Warp connection with a team, that connection keeps working for everyone else. If you want Warp gone entirely, disconnecting it once in Elaichi removes it from Claude, ChatGPT, Cursor and every other client at the same time.

## All 27 Warp tools

Every tool below is callable through https://api.elaichi.ai/mcp once Warp is connected, subject to the toolbox it is in and the restrictions on the caller.

- **List all Warp agents** (List). List available Warp agents (skills) that can be used to run tasks, discovered from accessible environments or a specific repository. Returns each agent's name and a variants array; each variant includes id, description, base_prompt, source (owner, name, skill_path, worker_host), environments, last_run_timestamp, and error.
- **List all Warp agent connected self hosted workers** (List). List currently connected self-hosted workers for the authenticated team in Warp. Returns: worker_host, connection_count, connected_at, last_seen_at. Worker presence is derived from websocket heartbeats and may be briefly stale.
- **List all Warp run transcripts** (List). Retrieve the raw conversation transcript for a warp agent run. The API issues a 302 redirect to a time-limited download URL containing the transcript content; no structured JSON body is returned on success. Required: run_id.
- **Create a Warp agent run** (Create). Spawn a Warp cloud agent run with a prompt and optional configuration. The agent is queued for execution and assigned a unique run ID. Returns: run_id, task_id, state, at_capacity. Required: prompt.
- **List all Warp agent runs** (List). List Warp agent runs with optional filtering by state, creator, model, date range, skill, and more. Results default to sort_by=updated_at and sort_order=desc. Returns: run_id, task_id, title, state, execution_location, prompt, created_at, updated_at, run_time, started_at, status_message, source, schedule, session_id, session_link, trigger_url, creator, executor, request_usage, agent_config,…
- **Get single Warp agent run by ID** (Get). Get detailed information about a specific Warp agent run by id, including the full prompt, session link, and resolved agent configuration. Returns: run_id, task_id, title, state, execution_location, prompt, created_at, updated_at, run_time, started_at, status_message, source, schedule, session_id, session_link, trigger_url, creator, executor, request_usage, agent_config, conversation_id,…
- **Warp agent runs cancel** (Cancel). Cancel a Warp agent run that is queued or in progress. Returns the run_id string of the cancelled run. Required: run_id. Cannot cancel runs already in a terminal state (SUCCEEDED, FAILED, ERROR, BLOCKED, CANCELLED); PENDING-state runs return 409.
- **List all Warp run timelines** (List). List chronological setup and lifecycle timeline events for a warp agent run. Returns an events array where each item includes event_uuid, run_id, execution_id, event_type, occurred_at, and payload. Required: run_id.
- **List all Warp run conversations** (List). Get normalized run conversation in Warp by id. Returns conversation_id, steps with descriptions, summaries, timestamps, and nested messages showing user and system interactions.
- **Create a Warp run followup** (Create). Submit a follow-up message to an existing warp agent run. The server transparently routes the message based on the run's current state (queued, actively running, or ended). Returns an empty 200 response on success. Required: run_id.
- **Get single Warp agent conversation by ID** (Get). Get a normalized conversation in Warp by id. Returns conversation_id, steps including step descriptions, timestamps, and messages with their content.
- **List all Warp agent schedules** (List). List all scheduled agents accessible to the authenticated user in Warp, sorted alphabetically by name. Returns: id, name, cron_schedule, enabled, prompt, last_spawn_error, agent_config, agent_uid, environment, created_at, updated_at, created_by, updated_by, history, scope.
- **Create a Warp agent schedule** (Create). Create a new scheduled agent in Warp that runs automatically on a cron expression. Returns the created agent including id, name, cron_schedule, enabled, prompt, agent_config, agent_uid, created_at, history, and scope. Required: name, cron_schedule.
- **Get single Warp agent schedule by ID** (Get). Get a scheduled agent by id in Warp, including its full configuration, run history, and next scheduled execution time. Returns: id, name, cron_schedule, enabled, prompt, last_spawn_error, agent_config, agent_uid, environment, created_at, updated_at, created_by, updated_by, history, scope. Required: id.
- **Update a Warp agent schedule by ID** (Update). Update an existing scheduled agent in Warp. Returns the updated agent including id, name, cron_schedule, enabled, prompt, agent_config, agent_uid, created_at, history, and scope. Required: id, name, cron_schedule, enabled.
- **Delete a Warp agent schedule by ID** (Delete). Delete a scheduled agent by id in Warp, stopping all future scheduled runs. Returns a 200 response on success. Required: id.
- **Warp agent schedules pause** (Pause). Pause a scheduled agent in Warp, preventing it from running until resumed. Returns the updated schedule object including id, name, cron_schedule, enabled, prompt, agent_config, agent_uid, created_at, and updated_at. Required: schedule_id.
- **Warp agent schedules resume** (Resume). Resume a paused scheduled agent in Warp so it begins running again according to its cron schedule. Returns the updated schedule object including id, name, cron_schedule, enabled, prompt, agent_config, agent_uid, created_at, and updated_at. Required: schedule_id.
- **List all Warp agent environments** (List). List cloud environments accessible to the authenticated principal in warp, including environments the caller owns, has been granted guest access to, or has accessed via link sharing. Returns: uid, config, last_updated, last_task_run_timestamp, last_task_created, setup_failed, scope, creator, last_editor.
- **List all Warp agent models** (List). List LLM models available to the authenticated user in Warp for agent runs. Returns per-model fields including id, display_name, provider, vision_supported, description, reasoning_level, and disable_reason (when the model is unavailable).
- **Get single Warp agent artifact by ID** (Get). Get a single Warp artifact by id. For downloadable file-like artifacts returns a time-limited signed download URL; for plan artifacts returns the current content inline. Returns: artifact_uid, artifact_type, created_at, data. Required: id.
- **List all Warp harness support transcripts** (List). Download the raw third-party harness transcript for the current task's conversation in Warp. Returns a 307 redirect to a signed download URL for the raw transcript file (e.g., claude_code.json). Only supported for non-Oz harness conversations; must be called from within a cloud agent execution environment that has an associated conversation.
- **List all Warp agent identities** (List). List all warp agent identities for the caller's team. Returns each agent's uid, name, available, description, prompt, environment_id, created_at, updated_at, base_model, skills, secrets, memory_stores, base_harness, inference_providers, and harness_auth_secrets.
- **Create a Warp agent identity** (Create). Create a new warp agent identity for the caller's team. Returns the created agent object including uid, name, available, created_at, updated_at, description, prompt, environment_id, base_model, secrets, skills, memory_stores, base_harness, inference_providers, and harness_auth_secrets. Required: name.
- **Update a Warp agent identity by ID** (Update). Update an existing warp agent identity by id. Returns the updated agent object including uid, name, available, created_at, updated_at, description, prompt, environment_id, base_model, secrets, skills, memory_stores, base_harness, inference_providers, and harness_auth_secrets. Required: id.
- **Delete a Warp agent identity by ID** (Delete). Delete a warp agent identity by id. All API keys associated with the agent are deleted atomically. Returns an empty 204 response on success. Required: id.
- **Get single Warp agent identity by ID** (Get). Retrieve a single Warp agent identity by id. Returns the full agent object including uid, name, available flag, created_at, updated_at, secrets, skills, memory_stores, inference_providers, base_model, base_harness, harness_auth_secrets, description, prompt, and environment_id. Required: id.