Rush docs

Send a request.
Get the work back, with proof.

An agent does the work in its own cloud sandbox, with a computer and a browser, and returns a pull request, a report, files or an answer, beside the full record of how it got there.

Run your first agent Browse use casesBase URLhttps://api.phoenix.so
View .mdAgent skill (SKILL.md)

04It comes back with proof

01 · REQUEST02 · COMPUTER03 · LIVE WORK04 · PROOFPOST /o/{org}/sessions{ "agent": "claude", "input": "Fix the flaky test", "resources": [...] }202 Acceptedqueuedallocatingrunningits own sandboxrepo checked outno other runevent: tool_usebun test checkoutevent: tool_result1 failed, 11 passedevent: thinkingthe test races its timerevent: tool_result12 passedcompletedbranchrush/fix-flaky…pr_url…/pull/412total_cost_usd0.37
Scene. One request, start to finish. Field and event names are the API's; the values are an example.

Three ways in, one API

04Each is the same kind of session

Your serveran API keyThe consolea teammate, signed inYour coding agentClaude Code, Codex, CursorRush APIapi.phoenix.soone routeSESSIONharnessclaudestatusrunningstarted_bywhoever started itrecordtranscript, result
Scene. Three ways to start a run, one API behind them, the same session at the end.

From your coding agent

Hand Rush a task without leaving Claude Code, Codex or Cursor. Add the MCP server once, or point any agent at the skill, and it starts runs as tool calls.

MCP server
claude mcp add --transport http rush https://api.phoenix.so/mcp --header "Authorization: Bearer $RUSH_API_KEY"

What people send

All 11 use cases

What comes back

Every run ends with its result beside the record of how it got there: the pull request and its checks, the files it saved, every command it ran and what it cost.

When the proof is missing, the run says so. It ends needs_review, never a false completed.

Results and proof
cloud.getrush.ai · run-af7b2434
A finished cloud run in the Rush console: a review agent ran the code of pull request #2500. Status completed, its model, turns, tool calls, tokens and cost, and the repository and pull request in the side panel.
A real run in the console: a review agent ran the code of pull request #2500, then reported back. Status, cost, tool calls and the PR sit beside the transcript.

How it works

01Runs and sessionsA session is one agent run. It moves through a fixed set of statuses, takes follow-ups between turns and keeps its sandbox while it is parked.02Projects and secretsA Project ties a repository, its secrets and its shared connections together inside your org. Sessions in it get all three without naming them.03Results and proofEvery run ends with a record you can check: the answer, the branch and pull request, the files it saved, its transcript and what it used.04Watch and steerFollow a run as it works, send it the next instruction, stop it, or take over its desktop. Every surface reads the same event stream.05IsolationEach run gets its own microVM, each org sees only its own resources, and secrets reach a run only in the form you chose. What that means in practice.06ModelsAPI key and template-agent runs use Rush's models, billed per token; your own runs, in the console or an app you signed in, use your own account.07AutomationsAn automation is a saved prompt in a Project with triggers that say when it runs: a GitHub event, a schedule, a webhook call or a Linear issue change.08Saved agentsSave a harness with its model, effort, instructions, skills and connections once, then start every run of it by sending its id as agent.09ConnectionsSign in to a tool or MCP server once, then name it on any run. Skills and resources come in by URL the same way.10WebhooksGive an API key an endpoint and every run it starts posts a signed event there when a turn ends, so you never hold a stream or poll.11Usage and billingRush meters compute while a run is awake and tokens on Rush's models. Every session records its tokens and cost; limits answer with a typed code.12MCP serverHand Rush a task from claude.ai, ChatGPT, Claude Code, Cursor or Codex: connect api.phoenix.so/mcp by signing in or with an API key, then start runs as tool calls.13Community agentsCompanies on Rush share saved agents they run. Add one and it becomes your own agent, running on your own apps, while its instructions stay with its maker.

Learn

Do it once, start to finish.

Lessons that walk a run from the first request to the pull request, from the console or your own code.

Start learning