Buster
Buster (YC W24), now Keystroke · Coding · Marketing
Buster is a YC-backed AI automation platform where teams build TypeScript agents and workflows with coding agents, then run them from one shared web app.
About Buster
What Is Buster
Buster is a Y Combinator W24 startup from Blake Rouse and Dallin Bentley. It began as an AI data analyst, then repositioned as a code-first AI automation platform marketed as an n8n alternative for coding agents. The pitch was simple: give teams the integrations and hosting of a visual tool like n8n, but keep the automation in ordinary TypeScript that a coding agent can read, test and change.
The work lives in your own repository. A workflow is an async function, inputs are Zod schemas, and Vitest covers the logic. Agents wrap Vercel's AI SDK and treat workflows, actions, other agents, MCP servers and integrations as tools. Once pushed, everything shows up in a hosted app with a chat interface, run history, credentials, sharing controls and access rules. That split matters: engineers keep a real codebase, and the rest of the company gets something they can actually use.
The catch is status. The buster.so site no longer serves a live page, and the team has moved on. Buster's founders rebuilt the idea as Keystroke, an open source agent platform with the same code-first approach and a broader feature set. Treat Buster as a discontinued product and evaluate Keystroke if you want the current version.
Getting Started
- Install the CLI and scaffold a local project. The command creates a working directory with an AGENTS.md file that explains the conventions to your coding agent.
- Open that folder in Claude Code, Cursor or Codex and describe the agent you want. The coding agent writes the TypeScript, adds tools and runs the tests.
- Define each workflow input with a Zod schema so the deployed form validates data before a run starts.
- Push the project to the platform, then check the generated chat UI, workflow view and logs.
- Invite teammates, attach credentials for each integration, and set who can run or edit what.
Product Information
A quick look at Buster's pricing, supported platforms, and performance.
Best for
The users, tasks, and scenarios where this tool fits best.
Users
- Engineering teams
- Operations teams
- Solo builders
Tasks
- Internal tools
- AI workflow automation
- Inbound triage
- Reporting
Scenarios
- Replacing a visual builder
- Letting non-developers use what engineers build
- Rebuilding an AI data analyst
Key features
TypeScript Workflows
Workflows are plain async TypeScript. A loop is a real loop, inputs use Zod schemas, and the whole project runs under Vitest. You get npm packages, type checks and diffs that a reviewer can read, which is the main reason teams pick this over a canvas.
Durable Managed Runtime
The hosted runtime keeps a run alive across errors, sleeps, approval steps and process restarts. Long jobs that wait on a human decision don't lose their place, and a failed step can retry without restarting the whole chain.
Agents With Real Tools
Agents wrap Vercel's AI SDK and can call workflows, actions, other agents, MCP servers and integrations as tools. They also carry memory, a persistent file system, web search and triggers, so an agent can hold context between conversations instead of starting cold each time.
MCP and Integration Support
MCP servers plug in as tools, alongside common services such as Slack, Postgres, PostHog and Metabase. Credentials are stored per integration and shared through the platform, so each agent doesn't need its own copy of a key.
Shared Web App
Deployed agents and workflows appear in a browser app with a chat UI, workflow visualizations, input forms, logs and sharing controls. Non-technical teammates can run a job or ask a question without cloning a repo or installing anything.
Built for Coding Agents
The CLI scaffolds a project with an AGENTS.md file, and the docs are written so tools like Claude Code, Cursor and Codex can search them. Your coding agent can build the system, run tests and push it up, which turns TypeScript agents into something a small team can maintain.
Pros and cons
Pros
- Automation stays in your own repository as readable TypeScript, so code review, tests and version control work the way they already do.
- The managed runtime handles durability, approvals and restarts, which saves weeks of infrastructure work for long-running jobs.
- Agents can use MCP servers, workflows, other agents and standard integrations as tools, so complex systems stay modular.
- The hosted app gives the whole company a usable front end with logs, credentials and permissions.
- The workflow is designed around coding agents, and the successor product is open source, so you can inspect the code before committing.
Cons
- You need TypeScript skills or a coding agent you trust. There's no drag and drop builder for non-developers.
- Buster itself is discontinued and its site is offline, so new users can't sign up for the original product.
- Pricing, roadmap and support now belong to Keystroke, which means the documented offer may change again.
- The successor launched in open alpha, so expect rough edges and breaking changes in parts of the platform.
Frequently asked questions
Buster is used to build internal AI agents and workflow automations in TypeScript, then run them for a team through a hosted web app. Typical jobs include Slack assistants, reporting pipelines and ticket triage.
Related content
Explore related tools, skills, and articles for Buster.
Buster Alternatives
Base Code
Base44 (Wix) · Coding · ProductivityBase Code is a shared cloud development environment from Base44. Teammates describe changes to an existing GitHub codebase in chat, preview them live and send them back as pull requests.
Google EmbeddingGemma 2
Google DeepMind · Writing · Image · CodingGoogle EmbeddingGemma 2 is a 740M-parameter open model that turns text, code, images, audio, and video frames into vectors in one 768-dimension space, running on phones and laptops.
GitGlow
GitGlow · Coding · ProductivityGitGlow is a local-first desktop Git client that lets independent developers review their own code and their AI agents' changes in one place before shipping.