TAI Labs
Free to join, free to read, and you can start your own room. Build browser agents that research companies, watch competitors and prep your interviews — and see every step they take. Public AI communities for builders, operators, engineers and leaders, plus city rooms in London, San Francisco and New York. Also home to the AI Career Accelerator and AI-Native Team Training for companies.
https://tailabs.ai/Connect straight to this server’s public endpoint.
https://tailabs.ai/mcpWe add this server to your workspace, walk through sign-in once, then open Studio with tools ready to run.
Last probed Sep 14, 2026 · tailabs.ai
32tools discovered
Showing 25 of 32 from the live probe.
Agent templates
What a TAI Labs browser agent can be built to do: the starting points, what each one asks for, and roughly what a run costs in credits. Answers without an account, so it is the right first call for someone deciding whether TAI Labs is worth signing up for. Use create_agent to save one once they are signed in.
My agents
The signed-in person's saved agents, plus any agent a teammate has shared with their current workspace: what each one is told to do, whether it runs on a schedule, and when it last ran. Use before running or creating one, so an existing agent is reused rather than duplicated — including one a colleague has already built. A shared agent can be run but not renamed, rescheduled or deleted, and running it spends the team's pooled credits.
Create an agent
Save a new agent for the signed-in person. The goal is the instruction it will be given each run, so write it as a complete task rather than a topic. Set a cadence to have it run on its own; leave it manual to run it only on request. Set visibility to 'shared' to let everyone in their team workspace run it too — that needs a team workspace, and it stays theirs to edit. If the job is behind a sign-in, pass a profileId from my_logins.
Reschedule, pause or share an agent
Change a saved agent's schedule, pause or resume it, or share it with the workspace. Only the fields you pass are changed. Ask which timezone the person means rather than guessing — an agent set in the wrong zone runs at the wrong time every day and looks broken. Name, goal and start page cannot be changed here: create a new agent and delete the old one.
Delete an agent
Permanently delete one of the signed-in person's saved agents. Its past runs are kept. Only the person who created an agent can delete it — a teammate's shared agent can be run but not removed. This cannot be undone, so confirm the agent's name with the person before calling it. To stop an agent without losing it, use update_agent with enabled false instead.
Run an agent
Run one of the signed-in person's saved agents now — or one a teammate has shared with their workspace — and return what it found. Runs cost credits and can take a minute or two; a shared agent spends the team's pooled credits rather than the runner's own. Use my_agents first to get the id.
Check an agent run
How one agent run is going, or how it ended: status, summary, error, steps used, credits charged and the step-by-step trail of what the agent did. Use it after run_agent reports a run is still working, with the run id it returned. Poll it rather than asking for a long wait.
My agent runs
The signed-in person's recent agent runs, newest first: what each one was asked to do, how it ended, how many steps it took and what it cost in credits. Pass an agentId to see one agent's own history. Use run_status for the full step trail of a single run.
My saved logins
The saved browser logins the signed-in person's agents can use: which site, which username, and a masked hint — never a password. Pass a profileId from here to create_agent when the agent has to get past a sign-in wall. Saving or changing a login is only possible in the TAI Labs web app.
My work
What the signed-in person has actually shipped: completed agent runs with their results, anything in flight right now, the saved agents standing ready, how many rooms they are in and whether their public profile is worth sharing. The same view as the Work page in TAI Labs.
Team work
What the signed-in person's workspace has shipped, person by person: who has shared agents, who has completed runs, and who has not started yet. Only answers inside a team workspace, and only for someone who belongs to it. Use it for 'how is my team doing with this' and 'who needs a nudge'.
My access
Resolve the signed-in user's TAI Labs access: workspaces, active workspace, plan, role, entitlements, restrictions and why.
My communities
The rooms the signed-in person is in, and the ones they could join. Use to answer questions about where a conversation is happening or which room something belongs in.
Join a room
Join one of the TAI Labs rooms the signed-in person can join. Get the slug from my_communities. Joining an open room is instant; an invite-only or paid room refuses with the reason. Safe to call on a room they are already in.
Threads in a room
The conversations the signed-in person can open inside one TAI Labs room: channels, cohort channels and DMs, each with its thread id, unread count and last message. Use to find the thread id needed by read_thread or send_message.
Read a conversation
Read messages in one TAI Labs conversation the signed-in person has access to, oldest first. Returns up to 40 per call with a cursor for older messages. Get thread ids from room_threads.
Send a message
Post a message into a TAI Labs conversation as the signed-in person. Optionally reply to another message, or mention people by user id. Get thread ids from room_threads and user ids from room_members. This is visible to other people — confirm the wording before sending.
Open a direct message
Get the thread id for a one-to-one conversation with someone in a TAI Labs room the signed-in person is also in, creating it if it does not exist yet. Pass the other person's user id from room_members or find_person. Then use read_thread or send_message with the returned thread id.
Members of a room
The people in one TAI Labs room the signed-in person belongs to: name, headline, level, points and user id. Use to find the user id needed by open_dm, or to answer questions about who is in a room. Returns nothing for a room the caller is not in.
Find a person
Search TAI Labs people by name or headline and return their user id, so they can be mentioned or DM'd. Pass a room slug to search that room's full roster; omit it to search the community directory. Never returns email addresses.
Room feed
Posts from the TAI Labs community feed: title, author, likes and comment count, newest first. Pass a room slug from my_communities to read one room, or omit it for everything the signed-in person can see. This is the posts feed — use room_threads and read_thread for the running chat in a room's channels.
Post to a room
Write a post to a TAI Labs room's feed as the signed-in person. Get the slug from my_communities. This is visible to other people — show the person the title and body and confirm the wording before posting. Use send_message instead for chat in a room's channels.
Upcoming events
The upcoming TAI Labs community events the signed-in person can see: what it is, when it starts, who is hosting, how many are coming, and whether they have RSVP'd. Use rsvp_event with the id to say they are coming.
RSVP to an event
Toggle the signed-in person's RSVP to a TAI Labs community event. Calling it once says they are coming; calling it again takes that back — so do not repeat it to be sure. Get the event id from community_events.
My credits
The signed-in person's AI credit wallet: plan, credits remaining, how many were included versus topped up, how much of this cycle is used, a breakdown of what spent them, recent transactions, and whether auto top-up will refill it. Use for any 'how many credits do I have' or 'where did my credits go' question.
Get your MCP into directories
A working endpoint is step one. Directory coverage is the coordinated launch across ChatGPT, Claude, Cursor, the MCP Registry, and community indexes.
Directory coverage for brandsFree to join, free to read, and you can start your own room. Build browser agents that research companies, watch competitors and prep your interviews — and see every step they take. Public AI communities for builders, operators, engineers and leaders, plus city rooms in London, San Francisco and New York. Also home to the AI Career Accelerator and AI-Native Team Training for companies.
Use the MCP endpoint listed on this page in your MCP client configuration. One-click install pills support Claude, Cursor, VS Code, and other hosts. Copy the remote MCP URL if your client needs a manual entry.
Operate TAI Labs? Verify ownership to take over this directory entry.
This server appears in the MCPBundles directory. Verify you operate it to take over the listing — name, description, logo, contact email, and skill content. We email a 6-digit code to a maintainer address your server publishes in /.well-known/security.txt or /.well-known/mcpbundles.json. Free, takes about a minute.
MCPBundles probed 32 tools on the live server. The tool list on this page reflects what was discovered at the last refresh — connect your client to see the full set available to your session.
TAI Labs may require signing in to the provider before tools can run. Connect through MCPBundles or your MCP client and complete any provider login when prompted.
MCPBundles is an independent platform built on the open Model Context Protocol standard. Not affiliated with Anthropic PBC or Claude.