Knowledge CenterAPI ReferencesDalil MCPClaude Skills

Dalil MCP

The Dalil MCP server exposes your workspace to any Model Context Protocol client: Claude Desktop, Claude Code, Cursor, Cline, Continue, Zed, and more. Your AI assistant gets native tools for records, sequences, workflows, pipelines, messaging, and meeting notes. Just add one URL and sign in with your Dalil workspace.

Connect

One URL. No API key.

https://app.usedalil.ai/mcp

Add this URL to any MCP client and it will open a Dalil sign-in when you connect. You authenticate with your logged-in workspace (standard MCP OAuth): no API keys to create, paste, or rotate.

Claude (claude.ai and Claude Desktop)

  1. Open Settings → Connectors
  2. Click Add custom connector
  3. Name it Dalil and paste https://app.usedalil.ai/mcp
  4. Click Connect and sign in with your Dalil workspace account

Claude Code

claude mcp add --transport http dalil https://app.usedalil.ai/mcp

# then inside Claude Code, authenticate with:
/mcp

ChatGPT

  1. Open Settings → Apps & Connectors (enable Developer mode under Advanced if you haven't)
  2. Click Create to add a custom connector
  3. Paste https://app.usedalil.ai/mcp as the MCP server URL, choose OAuth authentication
  4. Complete the Dalil sign-in when prompted

Cursor, Cline, Continue, Zed, and other MCP clients

Add the server to your client's MCP config; the client will run the OAuth sign-in on first use:

{
  "mcpServers": {
    "dalil": {
      "url": "https://app.usedalil.ai/mcp"
    }
  }
}

Headless and automation use

For scripts, CI, or agents with no browser to sign in with, the endpoint also accepts a Dalil API key as a bearer token:

{
  "mcpServers": {
    "dalil": {
      "url": "https://app.usedalil.ai/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}
💡 The MCP server is enabled per workspace. If your client can't connect, check that MCP is enabled for your plan; the same setup snippet lives in the app under Settings → AI → Set up MCP.

What your AI can do

Once connected, the client discovers the full tool catalog automatically. Every call runs in the context of the workspace you signed in with; a tool can never reach another workspace, and read-only tools are annotated so clients can treat them as safe.

Records & data model

ToolWhat it does
find_recordsREADSearch and filter records on any entity, standard or custom
create_recordWRITECreate a record on any entity
update_recordWRITEUpdate fields on a record
delete_recordWRITEDelete a record
list_object_metadataREADDiscover entities and their fields (names, types, options)
create_fieldWRITEAdd a custom field to an entity
update_fieldWRITEChange a custom field’s configuration
delete_fieldWRITERemove a custom field

Pipelines

ToolWhat it does
list_pipelinesREADDiscover the workspace’s pipelines and their stages
get_pipeline_recordsREADRead records in a pipeline
get_pipelines_by_recordREADFind which pipelines a record belongs to
create_pipeline_recordWRITEAdd a record to a pipeline
update_pipeline_recordWRITEUpdate a pipeline record (e.g. move its stage)
delete_pipeline_recordWRITERemove a record from a pipeline

Sequences

ToolWhat it does
list_sequencesREADList outreach sequences and their status
get_sequence_enrollmentsREADSee who is enrolled and where they are in the flow
create_sequence_stepWRITEAdd a step to a sequence flow
update_sequence_stepWRITEEdit a step’s content or settings
delete_sequence_stepWRITERemove a step
enroll_person_in_sequenceWRITEAdd a person to a campaign
pause_person_in_sequenceWRITEPause a person’s progression
unenroll_person_from_sequenceWRITERemove a person from a campaign

Workflows

ToolWhat it does
create_workflowWRITECreate a workflow with a draft version
get_workflowREADRead a version’s trigger and steps
set_workflow_triggerWRITEConfigure the trigger
create_workflow_version_stepWRITEAdd a step (auto-links to the previous one)
update_workflow_version_stepWRITEEdit a step
delete_workflow_version_stepWRITERemove a step
compute_step_output_schemaREADPreview a step’s output shape for later variables
create_workflow_draftWRITECreate an editable draft from a published version
activate_workflow_versionWRITEPublish a version
deactivate_workflow_versionWRITETurn a version off
run_workflow_versionWRITETrigger a manual run
get_workflow_runsREADRead run history and per-step results

Messaging & engagement

ToolWhat it does
get_messagesREADRead synced conversations (email, WhatsApp, LinkedIn)
get_timeline_threadsREADRead a record’s conversation timeline
SEND_EMAILWRITESend an email from a connected account
SEND_WHATSAPPWRITESend a WhatsApp message
SEND_LINKEDINWRITESend a LinkedIn message
draft_whatsapp_messageWRITESave a WhatsApp draft without sending
draft_linkedin_messageWRITESave a LinkedIn draft without sending
SEND_LINKEDIN_CONNECTIONWRITESend a LinkedIn connection request
VISIT_LINKEDIN_PROFILEWRITEVisit a LinkedIn profile

Meetings

ToolWhat it does
get_ai_meeting_notesREADRead AI meeting notes: summaries, transcripts, action items

Not exposed over MCP by design: raw HTTP requests and the sandboxed code interpreter, which stay inside Dalil's own agents.

What this unlocks

  • Chat-driven CRM: "Find opportunities in Negotiation with no activity in two weeks, and create a follow-up task for each owner."
  • Outreach operations: "Enroll everyone from yesterday's import with a verified email into the Q3 sequence, and pause anyone who replied."
  • Automation building: your assistant can build a workflow end to end: create it, set the trigger, add steps, activate, and run it.
  • Context for everything else: messages, meeting notes, and pipeline state become context your assistant reads before drafting or deciding.

MCP or Claude Skills?

They compose. MCP gives the assistant tools: live, workspace-scoped operations it can execute. Claude Skills give it knowledge: the conventions, formats, and gotchas of the Dalil API, so it uses those tools (or the raw REST/GraphQL API) correctly on the first try. For the best results in Claude Code, connect the MCP server and install the skills.