The agent runtime for shipping software
Assign the issue.
Drisp ships the rest.
Assign an issue to Drisp the way you'd assign a teammate. It moves the issue through every status itself — build, review, QA on a live browser — and runs the right workflow at each one. You set direction. It does the work.
Rate-limit the public API
Dark-mode polish pass
GraphQL schema cleanup
Flaky e2e on CI
Stripe webhook idempotency
Audit-log export
Webhook delivery retries
Cursor pagination
OAuth token refresh
Mobile nav regressions
Empty-state polish
Self-serve billing portal
Invite teammates
CSV import limits
Most issues ship hands-off, some loop a human in — and auto-merge is opt-in, never a silent default.
The agent is fine. The workflow around it is broken.
Coding agents already write good code. What breaks is everything around the run: the same prompt gives a different result every time, long runs drift with no checkpoint to recover from, and what works on one laptop never quite transfers to the team. You see the final diff — not the forty tool calls that produced it. And because the tooling assumes a human at the keyboard, engineering caps out at what one developer can babysit.
Drisp sits between you and the agent. A workflow defines the job declaratively — steps, constraints, exit criteria — and the runtime holds the agent to it. Every step is checkpointed and verifiable, every run leaves a full log. The same workflow runs on your laptop, in CI, or fanned out across runners on Drisp Cloud, and it gives your team the same result it gave you.
One issue,
walked to Done.
After you assign it, Drisp self-transitions the issue through each status and runs the right workflow at every stop. This is the whole loop, in order.
Drisp picks the issue up and builds.
Assignment is the trigger — no workflow to pick, no command to run. Drisp moves the issue to In Progress and starts fullstack-engineering, a ten-phase workflow that orients in the codebase, plans the change, and implements it to a green build. 11 plugins, 31 skills, one PR at the end.
The diff gets a real review.
The same workflow turns on its own work: it reads the diff, applies fixes, and holds the PR to the repo's gates before approving. The merge stays yours — auto-merge is a ceiling you opt into, never a silent default.
QA fans out in parallel.
Three real testing workflows — smoke, regression, and exploratory — run as parallel sessions against a live browser. The same fan-out works at issue scale: assign a sprint of tickets and a fleet of runners works them side by side.
Shipped, with the receipts.
PR merged, issue closed, and the full run log kept — every tool call, token, and test result. Most issues ship hands-off end to end; the ones that need judgment loop a human in and say exactly where.
Run it yourself,
or let us run it.
The Drisp CLI is the runtime. It sits on top of Claude Code or Codex and coordinates every step of every workflow. Run it on your own machines, or attach it to managed runners on Drisp Cloud.
Yours to run.
MIT-licensed. Install it on your laptop, run it in CI, or self-host it on your own infrastructure. Bring your own agent keys. Own your data.
- Free, forever
- Local-first dev loop
- Self-host on any container runtime
- Workflows install from the marketplace or from a local path
Managed runners.
Spin up a runner from the dashboard, attach the workflows it should run, and assign issues from anywhere. Same CLI underneath — Drisp Cloud handles the infra, schedules, queues, retries, checkpoints, and storage.
- Free to start — no card required
- Pay per run after that
- Triggers from your tracker, GitHub, cron, REST, Slack, or webhook
- SSO, audit log, role-based access for teams
Workflows are identical in both modes — whether you type drisp workflow install on your laptop or attach a workflow to a Cloud runner, the same install runs underneath.
Install workflows
like packages.
Browse community-maintained workflows or publish your own. Each one is scoped, tested, and version-pinned. Install with a single command on the CLI, or attach one to a Cloud runner.
playwright-automation
v0.0.21Writes, reviews, and stabilizes Playwright end-to-end tests from live product flows.
$ drisp workflow install playwright-automationrobot-automation
v0.0.21Generates Robot Framework suites with reusable keywords, structured fixtures, and CI-ready runs.
$ drisp workflow install robot-automationexploratory-testing
v0.0.14Runs charter-based exploratory sessions, captures evidence, and files actionable issues.
$ drisp workflow install exploratory-testingsmoke-testing
v0.0.14Defines and executes smoke suites against critical paths so breakages surface before release.
$ drisp workflow install smoke-testingregression-testing
v0.0.14Scopes impact-based regression runs, reruns the right areas, and reports deltas against the last release.
$ drisp workflow install regression-testingfullstack-engineering
v0.0.9Maps the codebase, implements the change, validates the result, and ships a PR for full-stack features and fixes.
$ drisp workflow install fullstack-engineeringThe runtime details.
Triggers
Assignment in your tracker is the primary trigger. Runs also start from cron, the REST API, Slack, or a webhook — the runtime doesn’t care who asked.
Observability
Every run streams its events: tool calls, tokens, cost, and artifacts. When something fails you see exactly where — at one agent or at a thousand.
Portability
Write a workflow once and run it on Claude Code or Codex. Drisp normalizes the event stream, so the same definition, plugins, and logs work on both.
Recovery
Every step is checkpointed. A run that dies resumes where it stopped instead of starting over — no silent drift on long jobs.
The CLI is free, forever. Drisp Cloud is free to start, then pay-per-run — cost scales with the work, not the seat count.
See pricing →Assign your first issue.
Install the CLI and run a workflow locally, or start free on Drisp Cloud and let a runner take the next ticket. Engineering that scales like a dial, one issue at a time.
$ npm install -g @drisp/cli