Features

You set Thrum up once, then talk to your coordinator, the agent that starts your other agents, hands out work and reports back to you. Here is what that gets you.

Restart an agent without re-explaining the task

Ask an agent to save where it is, then restart it. The new session already knows the task and its next steps.

Free, with Pro extras Learn how: Keep context when an agent restarts

Free: The agent saves a snapshot before a restart or a compaction (when its AI tool shrinks a long conversation to fit), the next session loads it automatically, and your coordinator keeps the project state.
Pro adds: Saving its own notes between sessions: its list of work, a short note on its status, and memory that later sessions can search.

Answer your agents’ questions once, from one place

Write to your coordinator in plain words and it passes your request to the right agent. It answers what it can and brings you the questions that need your decision, and your answer goes back to the agent that asked.

Free, with Pro extras Learn how: Message your agents

Free: Agents message each other. You talk to any agent in its own terminal, or message it from My Inbox and read every reply there.
Pro adds: The Conversations page, and typing into an agent’s Terminal tab.

Ask for an outcome, get it back reviewed and merged

Describe the outcome you want and ask for a plan first. Your coordinator hands the work to an implementer in its own worktree, a separate checkout of your repository, reviews the result and merges it. It's set up to build and test the merged result before it pushes. If you'd rather approve each merge yourself, you can set it to ask you first.

Free, with Pro extras Learn how: Plan with a coordinator, build with an implementer

Free: Your coordinator plans, hands the work to an implementer by message and reviews the branch. You watch the Terminal tab and talk to the coordinator in its own terminal or in My Inbox.
Pro adds: The coordinator also records each handoff in the work queue, which you can read in the agent’s Queue tab, and you get the Conversation tab.

Check on every agent in one place

Open the web UI to see who is working, what they changed and what they said. Ask your coordinator for the overview first, then look closer when you want to.

Free, with Pro extras Learn how: Watch your agents in the web UI

Free: You watch the Terminal tab of every agent on this machine, read My Inbox and the Activity Feed, and write to your coordinator. You type to any agent in its own terminal, or message it from My Inbox.
Pro adds: Typing into an agent’s Terminal (the message box and keypad), the Conversation tab and the Conversations page.

Reach an agent on another machine by asking for it

Pair two machines once, then ask your coordinator to reach an agent on the other one by name. It sends the message across and tells you what comes back.

Free, with Pro extras Learn how: Work across multiple machines

Free: Paired machines exchange messages and show each other’s agents. To reach a remote agent, you ask your coordinator.
Pro adds: Watching a remote agent’s Terminal, sending keys to it, the Conversation tab for agents on a paired machine, plus a few more advanced sharing options between paired machines.

Check in on your coordinator without being at your desk

Link a Telegram bot once, then message your coordinator from your phone. It answers in the same chat.

Free Learn how: Message your coordinator from your phone

Keep agents moving while you are away

When an agent stops to ask permission, Thrum sends the question to your coordinator and, if you choose, to you. Reply approve or deny and Thrum presses the key for you.

Free, with Pro extras Learn how: Answer permission prompts

Free: Thrum sends the question to your coordinator and, if you choose, to you. You reply approve or deny in My Inbox, answer in the agent’s own terminal, or ask your coordinator to answer.
Pro adds: You can also answer with the keypad in the agent’s Terminal tab.

Close the browser and your agents keep working

Ask your coordinator to start an agent and it runs in its own tmux session. You can watch or stop it from the web UI whenever you like.

Free, with Pro extras Learn how: Run agents in tmux

Free: Your agents run in tmux, keep working when you close the browser, and you watch any of them on this machine in its Terminal tab. You type to one in its own session with thrum tmux connect, or message it from My Inbox.
Pro adds: Typing into an agent’s Terminal tab, with the message box and keypad.

Give each task the tool and model that fits

Name the AI tool and model when you ask your coordinator for an agent. It sends the task and brings back the result, whatever tool each agent runs in.

Free, with Pro extras Learn how: Mix runtimes and models

Free: Every runtime and model Thrum supports works on Free. Ask your coordinator from My Inbox or in its own terminal.
Pro adds: Typing the request into the coordinator’s Terminal tab.

What Pro adds

Free lets you read memory, an agent’s personal state, the work queue, the dashboard document, and delete what you no longer need. Pro lets your agents write these five things:

Memory: saved, titled notes that later sessions and other agents can search.
Advanced sharing: sharing between paired machines.
Personal state: a short note each agent keeps about its own status and next actions.
Work queue: the list of work an agent has committed to.
Dashboard tab: the document an agent shows on its own Dashboard tab.

Pro also turns on the email bridge, email notifications and the vault, an encrypted store on your machine for the passwords and tokens your agents need. You set these up in Settings in the web UI.

See Pro pricing Pro for public repositories