AI agents

Your AI agents can work in Thicket

Connect an AI assistant with MCP, give a coding agent the thicket CLI and its agent skill, or build on the REST API and SDK. Each works with the access of the person who approves it.

Included on every plan, Free included.

Free plan. No card required. Includes a 30-day Pro trial.

$ npm install -g thicket-cli$ thicket auth login   Opening your browser to approve access… done.   Signed in as Sarah Chen (owner) in sunridge-labs$ thicket todos list --in "Website Redesign" --list Development   ○  Build the services overview              Marcus Reid   Oct 18   ○  Integrate the contact form with the API  Marcus Reid   Oct 20   ○  Add SEO meta tags and sitemap            Alex Torres   Oct 22$ thicket todo "Write the launch announcement" --in "Website Redesign" --assignee priya --due friday   Added to Website Redesign, assigned to Priya Sharma, due Friday.

What your agents can do in Thicket

The CLI and REST API reach the whole product: projects, to-dos, messages, docs and files, the calendar, boards, chat, check-ins, timesheets, reports and search. The MCP connector covers the everyday work in projects for assistants that don't run code. Changes appear in each project's activity, labeled with the app that made them.

Using ChatGPT, Claude Code or another AI assistant

Connect Thicket to the assistant you already use. Nothing to install.

Thicket in your AI assistant: ask what's due, what a thread decided or what's on the calendar. Add to-dos, move cards, post updates and log time. Sign in to Thicket, approve access, and your assistant works with the projects you can see.

Running a coding agent

Claude Code, Codex, Cursor and any other agent that can run shell commands.

Thicket from your terminal: thicket is the official command-line interface for Thicket. Work with projects, to-dos, messages, docs and files, boards, chat, search, reports, notifications and timesheets from a terminal or through an AI agent. Output is formatted for people in a terminal and JSON when piped, with suggested next commands, and thicket commands --json lists every command for an agent to read.

Agent skill for Thicket: thicket setup claude installs two skills for Claude Code and prints the two commands that add its plugin. For Cursor, Codex and other agents, thicket skill prints the same instructions, and the repository's install.md shows where each agent reads them.

Quick start

npm install -g thicket-cli && thicket auth login

Needs Node.js 20 or newer. Sign-in opens Thicket in your browser to approve access; add --scope read for a read-only connection.

Building your own agent or integration

Bring Thicket into your own tools, apps and integrations.

The REST API: every feature of the web app at /api/v1, the same routes the app runs on, described in an OpenAPI spec. Sign in with OAuth 2.1 and PKCE, or use a personal access token for a script.

The SDK and agent references: thicket-sdk is the TypeScript client, with retries and pagination. For agents that read before they act, the skill file, llms.txt and the Markdown editions of the help center describe the API in plain text.

# A personal access token from My settings > API tokens
auth="Authorization: Bearer $THICKET_TOKEN"
ua="YourAgent (you@example.com)"

# Post an update to a project's message board
curl -H "$auth" -A "$ua" -X POST \
  https://www.thickethq.com/api/v1/$ORG/recordings/$BOARD_ID/children \
  -H "Content-Type: application/json" \
  -d '{"type": "message", "title": "Week 42 summary", "content": "..."}'

# Check off a finished to-do
curl -H "$auth" -A "$ua" -X PUT \
  https://www.thickethq.com/api/v1/$ORG/recordings/$TODO_ID/completion

No AI inside, on purpose

There's no chatbot docked in the corner of Thicket, no AI summary button and no credits meter. You bring the AI you already use, such as ChatGPT or Claude Code, and decide what it can reach.

We put the work into a clean data model, an API that covers the whole product, and a workspace calm enough that people and agents can find what they need. When AI helps your team, it connects through an approved MCP connection or a credential you can revoke, and you stay in charge of what it can reach.

Questions about AI agents and Thicket

Do AI agents cost extra?

No. MCP, the REST API and the CLI are included on Free, Starter at $25/month and Pro at $49/month. A connection uses your account and adds nothing to your bill.

Whose permissions does my agent use?

Yours. A connection through the CLI or an MCP assistant acts as you, with your current access across your organizations and projects. When your access changes, its access changes with it.

How does an agent sign in?

The CLI and MCP assistants open Thicket in your browser so you can review and approve access, and their credentials refresh on their own. They're listed in My settings > Connected apps, where you can disconnect them. Scripts can use a personal access token from My settings > API tokens.

Can I give an agent read-only access?

Yes. Sign in the CLI with thicket auth login --scope read, create a read-only personal access token, or connect an MCP assistant that asks only to read. A read-only connection can look but can't change anything.

What happens to content an AI tool reads?

The AI app and its providers may process and keep what they read under your or your organization's agreement and settings with them, including with read-only access. Disconnecting stops later use of that connection; it doesn't undo actions or delete copies the provider already holds. Vesperion Gate doesn't use private Thicket customer content to train general-purpose AI models.

Put a project where your agent can help

Start a project, invite the team, and connect your agent through MCP, the CLI or the API. You can disconnect it anytime, and its work stays in the project's history.

Try Thicket Free

Free plan. No card required. Includes a 30-day Pro trial.

Read the API reference