Blog
Notes on running agents in production.
The API, the sandbox, and what we learned building the thing that runs them.
FeaturedGuidesOpenAIAPIOpenAI Agents API alternative: the same request, on any modelOpenAI's hosted Agents API runs OpenAI models on OpenAI's terms. If you need Claude, Kimi or your own key behind the same kind of session, here is what to use, and when to stay put.
GuidesClaudeManaged AgentsClaude Managed Agents is the best harness for Claude. Here's when you need something else.If you build only on Claude, use Claude Managed Agents. Four conditions where you need an alternative, quoted from Anthropic's docs, and where Gobare loses.
GuidesParallel AgentsWebhooksGit ArchitectureMulti-Agent WorkflowParallel Agents, One Codebase: One Agent per Ticket, One Branch per SessionHow to run a fleet of coding agents against a single repository without state corruption or git merge conflicts: isolated sessions, dedicated branches, webhook-driven events, and a single-writer push pipeline.
GuidesWebhooksAPIQuickstartRun a coding agent from your backend API: one POST, one signed webhookStart a coding agent on its own cloud machine with one request, get told by a signed webhook when it's done, collect the answer and files. Three snippets and a full runnable example.
Start building
The work your backend does, done by an agent.
One POST gives an agent its own computer — a workspace, a shell, a browser — and it runs until the work is done. Any model, on your own key.