Skip to main content
Glama
Cursor Cloud MCP

Cursor Cloud MCP

by BenItBuhner

Cursor Cloud MCP

Unofficial MCP server for Cursor Cloud Agents. Same sign-in and API handling as Cursor for Android. Not affiliated with Anysphere, Inc.

List agents, read transcripts, follow runs, and send follow-ups from any MCP client.

Install

Node 22.23.4+.

npm install
npm run build

Related MCP server: Paperclip AI Adapter for Antigravity CLI

Auth

Paste an API key from cursor.com/dashboard/api:

export CURSOR_API_KEY=key_...

Or sign in in the browser. That stores a key at ~/.config/cursor-cloud-mcp/credentials.json.

npm run login
npm run status
npm run logout

Run

npm start

~/.cursor/mcp.json:

{
  "mcpServers": {
    "cursor-cloud": {
      "command": "node",
      "args": ["/absolute/path/to/Cursor-Cloud-MCP/build/index.js"],
      "env": {
        "CURSOR_API_KEY": "${env:CURSOR_API_KEY}"
      }
    }
  }
}

npm run http serves POST /mcp on port 8091.

Set CURSOR_EXTENDED_MODE=1 for the same unofficial account endpoints Cursor for Android keeps behind Extended mode. Off by default. Call cursor_status first. The server describes its own tools.

License

MIT © 2026 Bennett Buhner

Available Tools

29 tools
agent_lifecycleC

Extended: pause | resume | wake | cancel_tool_call.

Inputs: bcId (required), action (required), runId, toolCallId.

archive_agentB

Archive (soft-delete). Idempotent.

Inputs: id (required).

cancel_runB

Cancel the active run (terminal, cannot resume).

Inputs: id (required), runId (required).

create_followup_runB

Send a follow-up prompt to an existing agent (new run).

Inputs: id (required), text (required), mode.

cursor_statusA

Check auth (v1/me) and report documented vs Extended-mode availability. Start here.

Inputs: no named parameters.

get_agentB

Durable metadata for one agent (repos, autoCreatePR, latestRunId).

Inputs: id (required).

get_artifact_urlB

15-minute presigned URL for one artifact path.

Inputs: id (required), path (required).

get_blobC

Extended: GetBlobForAgentKV one record blob (base64 blob_data).

Inputs: bcId (required), blobId (required).

get_conversationC

Full verbatim user/assistant transcript (v0/conversation). v1 has no equivalent.

Inputs: id (required), timeoutMs.

get_diff_detailsC

Extended: branch diff (GetBackgroundComposerDiffDetails).

Inputs: bcId (required).

get_prC

Extended: account view of any PR (SCMService/GetPullRequest).

Inputs: prUrl (required).

get_pr_diffC

Extended: PR file diffs (SCMService/GetPullRequestDiff).

Inputs: prUrl (required).

get_pr_statusC

Extended: checks + review decision + counts.

Inputs: prUrl (required).

get_runB

One run: status, result text, durationMs, pushed git branches/PRs.

Inputs: id (required), runId (required).

get_run_streamC

Fetch SSE run stream and summarize: status/assistant/thinking/tool_call/result events, tail text.

Inputs: id (required), runId (required), maxChars.

get_usageB

Token usage for an agent, total + per run.

Inputs: id (required), runId.

list_agentsC

List Cloud Agents (v1, newest first). Filter triage queue.

Inputs: limit, cursor, includeArchived.

list_agent_storesD

Extended: Project Context stores (ListAgentStores).

Inputs: pageToken.

list_agents_v0A

Legacy v0 list with repo/branch/PR/summary in one round-trip. Best for PR triage.

Inputs: limit, cursor.

list_artifactsC

List screenshots/videos/logs under artifacts/.

Inputs: id (required).

list_modelsA

Recommended model ids for create_agent.

Inputs: no named parameters.

list_pending_followupsD

Extended: account queue (ListPendingFollowups).

Inputs: bcId (required).

list_runsB

List runs (turns) for an agent, newest first.

Inputs: id (required), limit, cursor.

list_workspace_filesD

Extended: live VM workspace (ListWorkspaceFiles).

Inputs: bcId (required).

prewarm_transcriptD

Extended: StreamConversation PREWARM initial_state + blob ids. Raw record behind the transcript viewer.

Inputs: bcId (required).

project_lineageC

Extended: Project lineage = ListWorkersForManager + ListBackgroundComposerChildren. Workers, side chats, subagents.

Inputs: managerBcId (required).

read_store_fileC

Extended: ReadAgentStoreFile (Project Context / shared notes).

Inputs: storeId (required), relativePath (required).

start_side_chatD

Extended: StartSideChatBackgroundComposer.

Inputs: parentBcId (required), name.

steer_agentC

Extended: InjectBackgroundComposerContext (steer / queue message).

Inputs: bcId (required), text (required), expectedRunId.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.4/5.0

    Scored across 29 tools

    Disambiguation4/5

    Most tools target distinct resources and actions, and descriptions clarify differences. The main overlap is list_agents vs list_agents_v0 (legacy version), with minor potential confusion between get_run/get_run_stream and PR-related tools, but these are mostly resolvable by descriptions.

    Naming Consistency3/5

    Most names follow snake_case verb_noun conventions (list_*, get_*, create_*, cancel_*), but several are noun-first or state-oriented (cursor_status, project_lineage, agent_lifecycle, prewarm_transcript), creating a mixed but still readable pattern.

    Tool Count2/5

    29 tools is heavy for the server's apparent scope. The set includes legacy variants (list_agents_v0) and many extended endpoints, pushing well beyond the 3-15 range that tends to be well-scoped.

    Completeness2/5

    A critical gap is the absence of a create_agent tool, despite list_models explicitly referencing create_agent. There is also no update_agent or initial run creation, leaving core lifecycle operations incomplete despite extensive read and extended coverage.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers