Overview
Connect Autonomy to source control, deploys, issue trackers, chat and observability.
Integrations belong to the organization. An admin connects each tool once in Organization → Integrations, and each project then chooses how it uses it: which repository, deploy, Linear team, Jira project or Slack channel is its own. Per-project choices live in the project's Settings → Integrations.
Source control and CI
| Integration | What it does |
|---|---|
| GitHub App | Runs the project's test plan on every pull request, takes @autonomy requests in comments, and reports a GitHub check. |
| GitHub Actions | Starts runs from a workflow once the preview URL or build is ready. |
| GitLab CI | Starts web or mobile runs from a pipeline and gates the job on the result. |
| Jenkins | Starts web or mobile runs from a Jenkins job and gates the build. |
| Bitrise | Sends APKs or preview URLs from a Bitrise workflow. |
| CircleCI | Tests web previews and mobile builds from CircleCI jobs. |
Deploys
| Integration | What it does |
|---|---|
| Vercel | Runs a project's tests on every Vercel deployment and reports a Vercel deployment check. |
| Netlify | Runs a project's tests on every Netlify deploy and reports a GitHub check. |
| Expo EAS | Runs a project's mobile tests on every EAS build and reports a GitHub check. |
| Deploy webhooks | Runs a project's tests when your own CI, Railway or Coolify finishes a deploy. |
Issues and chat
| Integration | What it does |
|---|---|
| Linear and Jira | Turns an issue into one ticket, comments when runs see it again, and syncs the ticket's status back. |
| Slack | Posts run results and issues to each project's channel, and answers questions or starts runs when you mention it. |
| Notifications and webhooks | Sends run, issue and organization events by email, Slack and signed webhooks. |
Observability
| Integration | What it does |
|---|---|
| OpenTelemetry export | Exports a trace for each run, with one span per step, to any OTLP endpoint. |
| Datadog | Sends those traces to Datadog APM and finds failed steps there. |
Where to start
Start with whatever creates your preview or staging targets: often the GitHub App with Vercel or Netlify, or a deploy webhook for other hosts. Add Slack or a webhook so failures reach the team that owns the journey. Connect Linear or Jira once issues are worth tracking as tickets.
Integration rollout
- Choose the project the integration serves.
- Pick one test plan that proves a real user journey.
- Store Autonomy keys and secrets in your tool's secret store.
- Start one run by hand, then turn on automatic runs.
- Send failures and new issues to the team that owns the journey.
What integrations publish
Integrations publish a short result and a link to the run, not every screenshot. A reviewer should see whether the run passed and reach the full run page in one click. Only an allowlisted set of facts leaves Autonomy: never prompts, AI model details, cost or secrets.
Didn’t find it?