ChatGPT
MCP when available, or a guided API setup
Bring a call task from chat into a supervised workflow, then return the transcript and result to the conversation.
For AI agents and their operators
Your AI found a phone line.
ClawdCall gives ChatGPT, Codex, Claude, Gemini, Perplexity, Cursor, and OpenClaw a controlled path from one instruction to a real phone call — with the outcome returned to the workflow.
5 min included60 sec to connectHuman in control
Call completed · 01:42Table confirmed for 7:00 PM
Transcript and structured outcome returned.
statussummarytranscriptmetadatawebhook payload
Choose your setup
Start with your agent.
Select the agent you use to get its recommended connection path and a ready-to-use setup prompt.
Continue from ChatGPT
Take the answer into the real world.
Use ChatGPT as the setup copilot, then connect ClawdCall through MCP when your workspace supports it or through the API in a tool-capable runtime.
Best path: MCP or guided API setup
Help me set up ClawdCall so an agent can place one supervised phone call. Read https://clawdcall.com/agents.md and https://clawdcall.com/skill.md. First explain the safest connection path available in this ChatGPT workspace. Pause for me to enter credentials, verification codes, CAPTCHA, and approve the first call. Never ask me to paste an API key into chat.
One execution layer
A clear path from every starting point.
The chat or coding agent can change. The operating contract stays the same: understand the task, confirm permission, invoke the phone workflow, and return a useful result.
Codex
OpenAPI, MCP, and project-safe secrets
Let a coding agent inspect the interface, wire the call into your project, and verify the result without committing credentials.
Claude
MCP or skill-guided execution
Give Claude a precise operating guide for phone execution, consent checks, secure authentication, and outcome handoff.
Perplexity
Research-to-execution handoff
Move from an answer or recommendation into a documented setup path your execution environment can safely run.
Gemini
Guided setup plus API execution
Use Gemini to understand the workflow and prepare a call brief, then connect through a tool-capable runtime.
OpenClaw
Agent skill and webhook handoff
Add managed phone execution to an OpenClaw workflow and send the post-call status, transcript, and metadata back.
Cursor
OpenAPI and codebase integration
Ask your editor agent to implement a typed call workflow while keeping keys in protected runtime storage.
The handoff
One request in. One verified outcome out.
Describe the job
Give the agent a recipient, a specific call objective, the context it needs, and the outcome you want returned. Vague instructions should be clarified before any call begins.
Connect securely
The agent reads the skill, MCP card, or OpenAPI description. You complete credentials, OTP, CAPTCHA, and authorization. API keys stay in protected runtime storage — never in chat or source control.
Get the result
ClawdCall runs the phone workflow and sends back completion status, transcript, summary, metadata, and an optional webhook payload so the originating agent can continue the task.
Built for responsible execution
The agent does the work. You keep the authority.
Phone calls create real-world cost and contact real people. ClawdCall is designed around supervised onboarding, explicit objectives, consent-aware calling, and bounded access — not unconstrained dialing.
Read the calling and acceptable-use terms ↗Sensitive steps stay human
You enter credentials and verification codes, grant authorization, and approve the first call.
Calls need a valid context
Use account-holder, internal-team, or consent-based workflows where the recipient expects contact.
Usage remains visible
Five minutes are included to test. Agents should warn you before a workflow may create paid usage.
Results remain reviewable
Inspect the status, transcript, summary, and metadata before downstream automation acts on the result.
Technical resources
Everything an agent needs to evaluate ClawdCall.
Use these resources to understand the workflow, authentication, available tools, and safety requirements.
Common questions
Before your agent makes the call.
Can ChatGPT, Codex, Claude, Gemini, or Perplexity make a phone call with ClawdCall?
Yes, when the agent is connected to a tool-capable environment through MCP or the ClawdCall API. If the current chat cannot invoke tools directly, it can still prepare the call brief and guide the supervised setup for a connected runtime.
How does ClawdCall connect to my agent?
ClawdCall connects through its MCP server or API from a tool-capable runtime. Public agent instructions, OpenAPI metadata, and a supervised first-call guide explain the available setup paths.
What does the human need to do?
The human completes identity checks, verification codes, CAPTCHA, authorization, and first-call approval. Agents should never request that passwords, OTPs, or API keys be pasted into model-visible chat or committed to source control.
What comes back after a call?
A workflow can receive the call status, transcript, summary, call metadata, and a structured webhook payload so the originating agent or application can continue the task.
What should ClawdCall not be used for?
ClawdCall is not for bulk telemarketing, spam, political outreach, emergency services, or calls to people who have not consented to or would not reasonably expect the call.
How much does it cost to test?
The free plan includes five minutes. Agents should check the current pricing page and warn the user before starting workflows that may create paid usage.
Ready when your agent is
Turn “someone should call” into a completed task.
Start with one supervised call. Keep the transcript, outcome, and control.