MCP · headless

Give your AI the tools to do real work.

tam speaks MCP. Your AI assistant – Claude, for example – can read your workspace context, close tasks, and move the board right from its own interface, no switching to a tracker.

ChatGPTtam
Pull together every open task for the Q3 launch, find the blockers, and put together a plan for today
tam · mcp (working…)

Analyzed 12 workspace tasks, found 2 blockers and raised their priority

tam · Q3 Launch sprint
Payment error via SBPHigh
Launch the landing pageHigh
Grouped into the "Q3 Launch" sprint · board cards updated instantly
To do2
TA-88
Payment error via SBP
High
TA-91
Launch the landing page
High
In progress0
Review0
Done0

How this changes your day

Work in the chat you already use

No need to open the tracker. Tell your AI: "File a bug on the login flow" – the task shows up on the tam board instantly.

Full project context

The AI doesn't guess or ask you to repeat yourself – it reads your docs, tasks and sprints and already knows what you mean.

Safe by default

The AI can't go off script. Every action is scoped by permissions and lands in the audit history.

What your AI can do inside tam.

Full reference →

Here's what tam already does over MCP. Some tools are still private-preview only; the full list ships with the hosted release.

Working with tasks

Create tasks, bugs and stories, change statuses, assign owners, and leave comments.

Create tasksChange statusCommentsLink related tasks+ more

Working with knowledge

Search and read docs and wiki pages – the AI understands the project's context right away.

Workspace searchRead documentsExport to Markdown+ more

Connected to your code

Check PR status and link commits to tasks in GitHub.

Pull request statusCommit linking+ more

Sprints & backlog

Plan sprints, groom the backlog, and reorder priorities – no switching to a separate tool.

Create a sprintBacklog prioritiesTask estimates+ more

Automation

Set up rules that fire themselves on an event – no code required.

Event-based rulesAuto-assignmentNotifications+ more

Time tracking

Log time against a task and build workload reports right from the chat.

Log timePer-task reports+ more
Bring your own AI

Works with the AI you already use.

Claude connects today. ChatGPT, Codex and other clients are going through security and permissions testing before the hosted release.

ChatGPT
Claude
Claude Code
Codex
Any MCP client
Permissions stay in your control

Access is exactly what you allow – nothing more.

Every AI agent gets its own name and limited permissions, like a new person joining the team. The foundation already works; a self-serve panel to configure it is in private preview.

Release bot

Access configured manually

Active
Read tasks and boards
Change status and comment
Create new tasks
Delete tasks
Manage team members
Access billing
Permissions are limited from day one

What if it breaks something? The agent can't go beyond the permissions it was granted. You can revoke access at any time, with no impact on the rest of the team.

How to connect

1

Create an access

Set up a dedicated account for your AI agent with the access level it needs – like onboarding a new hire.

2

Copy the config

tam generates a ready-to-use connection file – nothing to assemble by hand.

3

Connect it to your AI

Add the config to Claude or another MCP client – the agent is ready to work.

t.Add connector
Name
tam
Server URL
Published when the global gateway launches
Add

tam is requesting access to your workspace's tasks, docs, and sprints

Allow

Give your AI access to your tasks today.

Connect tam over MCP through Claude and try next-generation project management.

Start freeRead the docs