Features

Your agents work on the record, not beside it.

Veyrd speaks the Model Context Protocol. An assistant reads the workspace, changes tasks and joins the discussion — and every call it makes is written down under its own name.

Join the waitlist Agents and API keys never count as seats

VEY-214 In review

Empty states for the mobile task list

Record · 9 entries

  • Kola moved this to In progress
  • Claude CodeOAuth
  • CursorAPI key
  • Nightly agentAPI key

Six tools

Enough to be useful. Not enough to wander.

The MCP server exposes one workspace, and exactly these calls against it.

  • readlist_taskslist and filter tasks
  • readsearch_tasksadvanced search
  • writecreate_taskcreate a task
  • writeupdate_taskupdate a task
  • writeadd_commentcomment on a task
  • readget_workspace_summaryheadline statistics

Two ways in

Sign in, or carry a key.

OAuth

Claude · Claude Code

  1. Add a custom connector or remote MCP server
  2. Leave the client id and secret blank
  3. Sign in and pick the workspace
no headers

API key

Cursor · your own scripts

  1. Create a named key in workspace settings
  2. The key is shown once, then stored hashed
  3. Revoke it without touching anything else
Authorization: Bearer tg_key_…

Attribution

Read the agents out on their own.

An agent is a participant, not a background process. Its edits sit in the same stream as everyone else's, and the stream can be filtered to either.

  • Kola moved this to In review14:30
  • Cursor linked pull request #21416:12
  • Claude Code added a QA checklist16:45
  • Claude Code commented with the test plan16:47
  • Ada marked a decision17:02

Early access

Open a workspace. Write the first line.

We’re opening Veyrd to teams in waves. Join the waitlist and we’ll email you when your workspace is ready.

See pricing