Integration · Developer tools

Connect Claude & ChatGPT to Skyvern

Ask your AI to start a Skyvern browser task, run one of your saved agents, or find out why last night's run failed. It can check a run's status, output and step-by-step timeline, read the screenshots, recordings and logs a run produced, schedule agents and manage browser sessions.

Connect Skyvern free

Free plan, no credit card. Takes about a minute.

Try asking

“Go to our supplier portal and download this month's invoice.”
The AI starts a Skyvern browser task with your prompt and starting URL and returns a run ID. The browser can take real actions on the site.
“Did the nightly price check agent succeed?”
It finds the agent, lists its recent runs and reads the status and output of the latest one.
“Why did run wr_123 fail?”
It reads the run's step-by-step timeline and lists the screenshots and logs it produced.

What your AI can do

Tasks and agents

Start browser tasks from a prompt, run, create and update saved agents, and retry a finished or failed agent run.

Runs and results

List runs, read a run's status, output and extracted data, follow its timeline, and open the screenshots, recordings, downloads and logs it produced.

Schedules and sessions

Schedule agents with a cron expression or interval, list schedules, and open, list and close browser sessions.

“Run the lead import agent every weekday at 7:00 Copenhagen time.”
It creates a schedule for the agent with a cron expression and time zone.
“Which saved logins can my tasks use?”
It lists your saved website credentials by name and type. The secrets themselves are not returned.
“Stop the run that is still going.”
It finds the running task and cancels it by its run ID.

24 tools for Skyvern

These tools are switched on when you connect. You can switch any of them off, or require your approval before it runs.

cancel_runDeletesStop a running task or agent run by `run_id`.
close_browser_sessionVisible to othersClose a browser session by `browser_session_id`.
create_agentChanges dataCreate a new agent from a workflow definition in `yaml_definition` or `json_definition`, optionally in a `folder_id`.
create_agent_scheduleChanges dataSchedule an agent to run automatically. Pass `workflow_permanent_id`, `timezone`, and either `cron_expression` or `interval_seconds`, plus optional `parameters`.
create_browser_sessionChanges dataOpen a persistent browser session that tasks can reuse, optionally starting at a `url` and with a `timeout`.
get_agentGet one agent by `workflow_permanent_id`, including its steps and input parameters; optional `version`.
get_artifact_contentGet the content of one run artifact by `artifact_id`.
get_browser_sessionGet one browser session by `browser_session_id`, including its status.
get_runGet a task or agent run by `run_id`, including its status, output and any extracted data.
get_run_timelineGet the step-by-step timeline of a run by `run_id`, showing what the browser agent did.
list_agent_runsList past runs of one agent by `workflow_id`, filtered by `status`, date range or `search_key`.
list_agent_schedulesList the schedules set up for one agent, by `workflow_permanent_id`.
list_agentsList saved agents (workflows), filtered by `search_key`, `status`, tags or `folder_id`. Use this to find an agent's ID.
list_browser_profilesList saved browser profiles (kept cookies and logins), optionally filtered by `search_key`.
list_browser_sessionsList the currently open browser sessions.
list_credentialsList the saved website logins (names and types only, not the secrets), optionally filtered by `search`. Use their IDs in tasks that need to log in.
list_foldersList the folders used to organize agents, optionally filtered by `search`.
list_run_artifactsList the files a run produced (screenshots, recordings, downloads, logs) by `run_id`, optionally filtered by `artifact_type`.
list_runsList recent task and agent runs, filtered by `status`, `run_type`, tags or a `search_key`. Paged with `page` and `limit`.
list_schedulesList all scheduled agent runs, optionally filtered by `search` or `status`.
retry_agent_runChanges dataRun a failed or finished agent run again by `workflow_run_id`, using the same inputs.
run_agentChanges dataRun a saved agent by `agent_id` with its input `parameters`. It drives a real browser and can take real actions on websites; returns a run ID to check with get_run.
run_taskChanges dataStart an AI browser task: Skyvern opens a browser and does what the `prompt` says, optionally starting at `url`. It can submit forms and take real actions on websites; returns a run ID to check with get_run.
update_agentChanges dataReplace an agent's workflow definition by `workflow_id`, passing the new `yaml_definition` or `json_definition`.

Set up in three steps

1
Pick the app
Create a free PipMCP account and choose Skyvern from the app list.
2
Paste your key
Paste your Skyvern API key. Sign in to Skyvern Cloud, open Settings (app.skyvern.com/settings) and copy the API key. The key is scoped to your organization.
3
Add the link to your AI
You get a personal MCP link. Add it to Claude, ChatGPT or Cursor:
1. Click your name, then Settings › Connectors › Add custom connector.
2. Paste your link as the Remote MCP server URL.
3. Switch it on from the + menu in a chat.

Questions

What can the AI do in Skyvern?

The AI can list and read agents, runs, run timelines, run artifacts, schedules, folders, browser sessions, browser profiles and saved login names. It can start browser tasks, run, create and update agents, retry and cancel runs, schedule agents and open or close browser sessions.

Does the AI see my Skyvern API key?

No. Your Skyvern API key is encrypted at rest and never shown to the AI. After you save it, it is not shown again, not even to you. The AI only sees the results of the tools it calls.

Can I control what the AI is allowed to do?

Yes. You choose which tools are switched on, so you can start by only reading runs and results. Tasks and agent runs drive a real browser that can submit forms and take real actions on websites, so starting them can require your approval, as can cancelling a run. Every tool call is logged.

Does it work with ChatGPT?

Yes. In ChatGPT go to Settings › Apps & Connectors › Advanced and turn on Developer mode, then add your PipMCP link. Developer mode needs a paid ChatGPT plan: Plus, Pro, Business or Enterprise. The same link also works in Claude (Settings › Connectors › Add custom connector), Cursor and other MCP clients.

What does it cost?

PipMCP has a free plan with no credit card required. Paid plans bill per completed task. You also need your own Skyvern account. Anything Skyvern charges for runs is between you and Skyvern.

How do I get the result of a task?

Starting a task or agent returns a run ID right away while the browser works. The AI then uses that ID to check the run's status, output and extracted data, and can list the screenshots, recordings and downloads once it is done.