Autonomy Docs
AI Assistants

Available Actions

What an MCP client can do with Autonomy plans, runs, traces, and environments.

MCP exposes Autonomy resources to a connected client. Some clients show friendly labels, but the underlying tool names are stable and useful when you are writing agent instructions.

Available tools

ToolPurpose
autonomy_list_test_plansFind reusable plans in the authorized organization. Filter by web, api, ios, or android.
autonomy_get_test_planRead one plan before running it or explaining existing coverage.
autonomy_generate_test_planDraft a plan from product intent, acceptance criteria, a ticket, or a PR summary.
autonomy_trigger_test_runQueue a run for an existing plan using a saved environment or explicit target fields.
autonomy_list_runsFind recent run status before triggering duplicate work.
autonomy_get_runInspect run summary, agent logs, and evidence links.
autonomy_get_run_tracesFetch trace detail for failure diagnosis.
autonomy_list_environmentsFind configured staging, preview, API, iOS, and Android targets.

Trigger targets

autonomy_trigger_test_run accepts platforms with web, api, ios, and android. Use environmentSlug when the target is already saved in Autonomy. Use targets when the agent or CI job has a fresh deployment or artifact.

PlatformTarget fields
Webtargets.web.baseUrl
APItargets.api.baseUrl, optional targets.api.runtimeValues
iOStargets.ios.artifactUrl or scanned targets.ios.storageId, plus targets.ios.bundleId
Androidtargets.android.artifactUrl or scanned targets.android.storageId, plus targets.android.packageName

Add branch, pr, and deployment.commitSha when the run should attach back to a pull request or commit.

Agent recipes

Check existing coverage

Codetext
Use Autonomy MCP to list test plans and recent runs for this project.
If a relevant run already exists for this branch or PR, summarize its status and evidence before triggering anything new.

Run a PR preview

Codetext
Use Autonomy MCP to trigger the relevant smoke plan against this preview URL.
Pass the branch, PR number, and commit SHA. Return the run id and evidence link when available.

Diagnose a failure

Codetext
Fetch the Autonomy run and run traces before proposing a fix.
Cite the failing step, trace detail, and evidence link that support your diagnosis.

Generate missing coverage

Codetext
If no existing Autonomy plan covers this acceptance criteria, generate one focused plan.
Prefer deterministic assertions and include API checks when UI state depends on backend state.

Scope

These actions work with Autonomy plans, runs, targets, traces, and evidence. They do not expose raw browser clicks, native device automation, or arbitrary network execution over MCP.

Good agent behavior

An MCP client should inspect existing plans and recent runs before triggering new work. When a run fails, the client should cite the evidence link or trace detail it used, rather than summarizing from memory. After a fix, rerun the same plan against the same target before broadening the scope.

On this page