For developers

The MCP-native CRM.

104 tools. 409 endpoints. Built so AI agents can drive your CRM the same way your team does — with permissions, audit logs, and dry-run safety.

{
  "mcpServers": {
    "conduyt": {
      "command": "npx",
      "args": ["tsx", "src/index.ts"],
      "cwd": "~/path/to/conduyt-mcp",
      "env": {
        "CONDUYT_API_URL": "https://conduyt.app",
        "CONDUYT_API_KEY": "cdy_your_key_here"
      }
    }
  }
}

Drop that into Claude Desktop. Restart. Done.

Get an API key → Read the docs

30-second primer

MCP, briefly.

Model Context Protocol is an open standard from Anthropic that lets AI models connect to external tools. Think of it like USB for AI: any MCP-compatible client can plug into any MCP server. Claude Desktop, Claude Code, Cursor, custom agents — they all speak it.

Conduyt ships with a native MCP server. 104 tools across 20 modules. Drop in your API key and your AI client has full CRM access through a typed, validated, audited interface.

If you’d rather not use MCP, the REST API works the same way. Same auth, same permissions, same audit log. More on that below.

Read the MCP spec →

Tool catalog

Everything your CRM does, accessible to your AI.

20 modules, 104 tools, one consistent naming convention (conduyt_<module>_<action>). Grouped by domain for navigation.

Contacts 8 tools
conduyt_contacts_list, conduyt_contacts_get, conduyt_contacts_create, conduyt_contacts_update, conduyt_contacts_delete, conduyt_contacts_merge, conduyt_contacts_enrich, conduyt_contacts_summarize
Deals 7 tools
conduyt_deals_list, conduyt_deals_get, conduyt_deals_create, conduyt_deals_update, conduyt_deals_bulk_assign, conduyt_deals_history, conduyt_deals_forecast
Pipelines 4 tools
conduyt_pipelines_list, conduyt_pipelines_create, conduyt_pipelines_update, conduyt_pipelines_analytics
Tasks 5 tools
conduyt_tasks_list, conduyt_tasks_create, conduyt_tasks_update, conduyt_tasks_complete, conduyt_tasks_calendar
Companies 5 tools
conduyt_companies_list, conduyt_companies_get, conduyt_companies_create, conduyt_companies_update, conduyt_companies_delete
Activities & Tags 6 tools
conduyt_activities_list, conduyt_activities_log, conduyt_tags_list, conduyt_tags_apply, conduyt_tags_remove, conduyt_tags_bulk_update
Users 6 tools
conduyt_users_list, conduyt_users_create, conduyt_users_update, conduyt_users_deactivate, conduyt_users_reactivate, conduyt_users_stats
Automations 6 tools
conduyt_automations_schema, conduyt_automations_validate, conduyt_automations_resolve, conduyt_automations_import, conduyt_automations_dry_run, conduyt_automations_publish
Email Sequences 4 tools
conduyt_email_sequences_list, conduyt_email_sequences_create, conduyt_email_sequences_enroll, conduyt_email_sequences_unenroll
Messaging 3 tools
conduyt_send_email, conduyt_send_sms, conduyt_compose_email
Calendar & Appointments 4 tools
conduyt_calendar_availability, conduyt_appointments_list, conduyt_appointments_create, conduyt_appointments_cancel
Forms 4 tools
conduyt_forms_list, conduyt_forms_create, conduyt_forms_submissions, conduyt_forms_update
Invoices 5 tools
conduyt_invoices_list, conduyt_invoices_create, conduyt_invoices_send, conduyt_invoices_payments, conduyt_invoices_payment_link
Webhooks 4 tools
conduyt_webhooks_list, conduyt_webhooks_create, conduyt_webhooks_update, conduyt_webhooks_delete
Drip Campaigns 5 tools
conduyt_drip_list, conduyt_drip_create, conduyt_drip_pause, conduyt_drip_resume, conduyt_drip_stop
Lead Scoring 3 tools
conduyt_lead_scoring_rules, conduyt_lead_scoring_update, conduyt_lead_scoring_recalculate
Products & Templates 5 tools
conduyt_products_list, conduyt_products_create, conduyt_templates_list, conduyt_templates_create, conduyt_templates_apply
Smart Lists & Custom Fields 5 tools
conduyt_smart_lists_list, conduyt_smart_lists_create, conduyt_custom_fields_list, conduyt_custom_fields_create, conduyt_custom_fields_update
Bulk & Dashboard 5 tools
conduyt_bulk_ai_import, conduyt_bulk_export_contacts, conduyt_bulk_export_deals, conduyt_dashboard_summary, conduyt_notifications_list
AI & Discovery 10 tools
conduyt_query_insights, conduyt_global_search, conduyt_api_catalog, conduyt_permission_introspect, conduyt_get_dashboard, conduyt_compose_email, conduyt_enrich_contact, conduyt_summarize_contact, conduyt_smart_lookup, conduyt_audit_log_query

Get started in minutes

Pick your client.

Claude Desktop

  1. Install Claude Desktop from claude.ai
  2. Get a Conduyt API key from your account settings
  3. Get the conduyt-mcp package from your account dashboard (available on request during the trial)
  4. Add the JSON config snippet to your Claude Desktop config
  5. Restart Claude Desktop
  6. Ask Claude: “How many open deals do we have?”

Claude calls conduyt_deals_list and answers in plain English.

Claude Code

  1. Install Claude Code via npm
  2. Get a Conduyt API key
  3. Add the same config to your MCP config file
  4. Run claude in your terminal
  5. Same 104 tools, available in your terminal session

AI-driven CRM ops without leaving your shell.

Anything else

  1. Install any MCP-compatible client (Cursor, Continue, custom agent)
  2. Point its MCP config at our server
  3. The 104 tools are exposed automatically
  4. No SDK required. No vendor lock-in.

Whatever speaks MCP, works.

Not using MCP?

The REST API works the same way.

Not every team is on MCP yet. Some are still on Zapier, n8n, or rolling custom agents against raw HTTP. We support those, too.

The REST API mirrors the MCP server one-to-one. Same auth (Bearer token, cdy_ prefix). Same permissions (scoped keys, per-domain access). Same audit log. Same 409 endpoints across 30 domains.

# Pull the dashboard summary
curl https://conduyt.app/api/v1/dashboard/summary \
  -H "Authorization: Bearer cdy_your_key_here" \
  -H "Content-Type: application/json"

Or call our schema-discovery endpoint, which returns the full catalog in one response:

curl https://conduyt.app/api/v1/schema/api-catalog \
  -H "Authorization: Bearer cdy_your_key_here"

That last one is what your AI agent should call first. It’s a self-describing index of every endpoint, parameter, and return type. Your agent reads it, learns the CRM, then drives.

Guardrails, not gates

AI access without the usual nightmares.

Open API access without controls is what makes a CTO refuse the meeting. So we built six primitives specifically for AI agents:

Scoped API keys

Read-only, write, or admin tiers. Restrict to specific modules (e.g., contacts only, no user management). Per-IP allowlisting available.

Dry-run mode

Add ?dry_run=true to any mutation. Returns what would happen without actually doing it. Perfect for letting your AI propose changes before you approve.

Sandbox flag

Run a full transaction inside a Postgres rollback. Use for “what if” simulations or AI experimentation. Nothing persists.

Confirmation tokens

Destructive actions (delete, bulk reassign) require a 5-minute-TTL handshake. Your AI proposes, you confirm with a separate API call.

AI action budgets

Default is 120 requests per minute per API key, enforced via Upstash at the Edge middleware layer. Hits the limit, gets a 429 with X-RateLimit-Reset.

Full audit trail

Every API call logged with who (key ID), what (endpoint + payload), when, where (IP). 90-day retention. Exportable as JSON or CSV for your SIEM.

See all the AI safety primitives in detail →

Your stack. Your AI. Your call.
Get an API key → Request MCP access