Give an AI agent eyes on a live website or app. The agent browses it, interacts with it, and tells you whether a given task or check passed. Works on localhost or any URL. Use for visual QA, flow validation, regression checks, or anything that needs a real browser to verify.

LOCALHOST SUPPORT: Pass any localhost URL (e.g. http://localhost:3000) and it Just Works. A secure tunnel is automatically created so the remote browser can reach your local dev server — no manual ngrok setup, no port forwarding, no config.

SCOPE PER CALL: Keep each call to ONE focused check — a single page or a short interaction on a single screen (login, submit a form, verify a heading). For anything spanning multiple pages or long multi-step flows, split into SEPARATE calls — the remote browser agent has a ~25-step internal budget per call, and long single calls risk client-side timeouts. Example: instead of "log in, then go to settings, then update profile, then verify," make three calls: (1) log in & verify dashboard, (2) update settings, (3) verify profile change.

Trigger a browser-agent crawl of a web app to build the project's knowledge graph. The crawl systematically explores pages, UI states, and navigation flows, then populates the backend's knowledge graph so future evaluations and tests have context about the app.

LOCALHOST SUPPORT: Pass any localhost URL (e.g. http://localhost:3000) and it Just Works. A secure tunnel is automatically created so the remote browser can reach your local dev server.

WHEN TO USE: after a significant new feature, a new environment, or when onboarding a project. NOT for per-change verification — use check_app_in_browser for that.

SCOPE: one crawl per call against one URL. The crawl is long-running (minutes to tens of minutes depending on app size) and populates backend state asynchronously; the tool returns the execution status once the workflow completes. This does NOT return pass/fail — it returns executionId + status + outcome.

Probe one or more URLs and return their rendered state — screenshot, page metadata (title/finalUrl/statusCode/loadTimeMs), structured console errors, and per-URL network summary (refetch loops collapse into one row by origin+pathname).

WHEN TO USE: "did I just break /settings?" / "smoke-test these 5 routes after my refactor" / "what's actually rendering at /dashboard?" — fast (<10s for 1 URL, <25s for 20), no LLM cost, no agent loop.

NOT FOR: scenario verification (sign in → click X → assert Y), interaction (clicks, form fills, scrolls), or anything requiring agent decisions. Use check_app_in_browser for those.

LOCALHOST SUPPORT: any localhost URL is auto-tunneled. Pre-flight TCP probe fails fast (<2s) if the dev server isn't listening.

BATCH MODE: pass up to 20 targets in one call to share browser session + tunnel — dramatically faster than firing parallel single-URL probes (one execution unit, not N). Per-URL waitForSelector / waitForLoadState / timeoutMs override defaults.

A single failed target's error appears in result.error without failing the whole batch — the other results stay valid.

Search or look up projects.

Two modes:

• uuid mode: pass {"uuid": ""} → returns that project with the curated detail view (uuid, name, slug, platform, repoName, description, status, language, framework, timestamp, lastMod), or isError:true NotFound.

• filter mode: omit uuid, optionally pass {"q": "", "page": 1, "pageSize": 20} → returns a paginated list of summaries (uuid, name, slug, repoName).

Response shape is always {filter, pageInfo, projects[]}. uuid mode returns exactly one project; filter mode returns summaries.

Search or look up environments, with credentials expanded inline per environment.

Two modes:

• uuid mode: {"uuid": ""} → single env with full detail + its credentials. NotFound if the uuid doesn't exist.

• filter mode: omit uuid, optionally {"q": "", "projectUuid", "page", "pageSize"} → paginated envs, each with its credentials.

Project resolution: if projectUuid is omitted, the current git repo's origin is auto-resolved to a DebuggAI project. Returns {error:"NoProjectResolved", environments:[]} if neither is available.

Credentials are returned inline per env as {uuid, label, username, role}. Password is NEVER returned — the handler defensively strips it regardless of what the service layer provides.

Response: {project, filter, pageInfo, environments[]} — each environment includes a credentials[] array.

Create a new environment under a DebuggAI project. Both name and url are required (backend rejects standard environments without a URL). Optional description. Defaults to the project resolved from the current git repo; pass projectUuid to target a different project (get UUIDs via search_projects).

OPTIONAL credentials seed: pass credentials: [{label, username, password, role?}] to create login credentials alongside the environment in a single call. Each cred is created best-effort; failures go to credentialWarnings without blocking env creation. Passwords are write-only and NEVER returned.

Returns the created environment's uuid (and the seeded credentials, if any). Reference the env uuid when running check_app_in_browser.

Patch an environment by UUID. Updates fields and/or manages credentials in a single call.

ENVIRONMENT FIELDS (all optional — only specified fields change):

• name, url, description

CREDENTIAL MANAGEMENT:

• addCredentials: [{label, username, password, role?}] — add one or more login credentials to this environment

• updateCredentials: [{uuid, label?, username?, password?, role?}] — patch existing credentials by UUID

• removeCredentialIds: ["", ...] — delete credentials by UUID

Operations run in order: remove → update → add. All credential ops are best-effort — failures go to credentialWarnings without blocking the rest. Passwords are write-only and NEVER returned in responses.

Returns {updated, environment, addedCredentials?, updatedCredentials?, removedCredentialIds?, credentialWarnings?}. Returns isError:true with NotFound when the env uuid doesn't exist.

Delete an environment by UUID. Returns {deleted: true, uuid}. Destructive — cascades per backend behavior (credentials under the env are typically removed). Defaults to the project resolved from the current git repo; pass projectUuid to target a different project. Returns isError:true with NotFound when the uuid doesn't exist or was already deleted.

Patch a project by UUID. Optional fields: name, description. Returns {updated:true, project:{...simplified resource}}. Returns isError:true + NotFound when uuid doesn't exist.

Delete a project by UUID. Returns {deleted:true, uuid}. DESTRUCTIVE — removes the project and its associated environments, credentials, and test history. No undo. Returns isError:true + NotFound when already deleted or uuid doesn't exist.

Search or look up workflow executions (history of check_app_in_browser, trigger_crawl, and other workflow runs).

Two modes:

• uuid mode: {"uuid": ""} → single execution with FULL detail including nodeExecutions, state, errorInfo. NotFound if the uuid doesn't exist.

• filter mode: {"status": "completed"|"running"|"failed"|"cancelled", "projectUuid": "...", "page", "pageSize"} → paginated summaries.

Response shape: {filter, pageInfo, executions[]}. Summary items have outcome/status/durationMs/timestamps; uuid-mode items additionally have nodeExecutions + state + errorInfo.

Create a new DebuggAI project. Required: name, platform (e.g. "web"), plus a team and a repo. Team and repo each accept EITHER a UUID or a name: pass teamUuid OR teamName, and repoUuid OR repoName. Name resolution does a backend search with case-insensitive exact match (returns AmbiguousMatch with candidates on multiple hits, NotFound on no hit). The repo must be GitHub-linked to the account. Returns {created: true, project: {uuid, name, slug, platform, repoName, ...}}.

Create a named test suite for a project. A test suite is a collection of test cases that can be run together. Requires name, description, and a project identifier (projectUuid or projectName). Returns {uuid, name, description, runStatus, testsCount}.

List and search test suites for a project. Returns paginated results with suite status, test counts, pass rates, and last run timestamps. Requires a project identifier (projectUuid or projectName). Optional: search text filter, page, pageSize (1-100, default 20).

Disable (soft-delete) a test suite. The suite and its tests are hidden from default list queries but not permanently removed. Accepts suiteUuid directly, or suiteName + project identifier for name-based lookup. Returns {deleted: true, suiteUuid}.

Create an individual test case and assign it to a test suite. The test is NOT automatically executed. Requires name, description, agentTaskDescription (the AI agent's goal), and suite + project identifiers. Optional: relativeUrl (must start with "/") and maxSteps (1-100). Returns {uuid, name, description, agentTaskDescription, suite, project, runCount}.

Update a test case's name, description, or agentTaskDescription. Requires testUuid. At least one of name, description, or agentTaskDescription must be provided. Returns the updated test case.

Disable (soft-delete) a test case. The test is hidden from default list queries but not permanently removed. Requires testUuid. Returns {deleted: true, testUuid}.

Trigger all test cases in a suite to run asynchronously. Accepts suiteUuid directly, or suiteName + project identifier. Optional: targetUrl to override the default test target. Returns {suiteUuid, runStatus, testsTriggered, note}. Use get_test_suite_results to poll for results.

Fetch a test suite with full per-test results. Returns suite-level status (NEVER_RUN, PENDING, RUNNING, COMPLETED, ERROR), pass rate, last run timestamp, and per-test outcomes (PASS, FAIL, ERROR, TIMEOUT, etc.) with execution times. Accepts suiteUuid directly or suiteName + project identifier.