MCP tools reference

This is the real tool catalog registered by the MaxoPerf MCP server — every entry here is something you can call today over POST https://app.maxoperf.com/mcp. Read tools default to response_format: "concise"; pass "detailed" for the full payload. Tools marked write need a non-read-only session; destructive ones ask for confirmation in Loadrigo and should get the same care from any agent driving them.

Context

Tool	Args	Returns
`whoami`	(none)	Account identity + default workspace.
`list_workspaces`	(none)	Workspaces visible to the account.
`set_active_workspace`	`{ workspace_id }`	Sets the active workspace for the rest of the session.

Projects

Tool	Args	Returns
`list_projects`	`{ workspace_id? }`	Projects visible in the account, with `canEdit`/`canDelete` flags.
`create_project` (write)	`{ workspace_id, name }`	New project.

Tests

Tool	Args	Returns
`list_tests`	`{ project_id?, workspace_id?, type?: 'single'\|'vario'\|'all', response_format? }`	Test list.
`get_test`	`{ test_id, response_format? }`	One test, including its validation summary.
`create_test` (write)	`{ project_id, name, type?: 'single'\|'vario', engine_kind? }`	New test shell — attach the entrypoint script next with `upload_test_file`.
`get_test_overview`	`{ test_id }`	Run-history overview (the console’s Overview tab).

Test files

Tool	Args	Returns
`upload_test_file` (write)	`{ test_id, filename, content? \| local_path?, role: 'entrypoint'\|'test_asset', content_type? }` — exactly one of `content` (inline text) or `local_path`	`{ fileId, uploadState, sizeBytes }`. One `entrypoint` per test; everything else is a `test_asset`.
`list_test_files`	`{ test_id }`	`{ files: [{ fileId, filename, fileRole, uploadState, sizeBytes }] }`
`download_test_file`	`{ test_id, file_id }`	`{ downloadUrl, expiresInSeconds, filename }` — a short-lived presigned URL, not raw bytes.

Runs

Tool	Args	Returns
`start_run` (write)	`{ test_id, engine_kind?, recording_layout?: 'tiled'\|'solo' }`	`{ runId, status }`. Idempotent — an identical retry reuses the run.
`get_run_status`	`{ run_id, response_format? }`	Lifecycle status: `queued \| allocating \| starting \| running \| stopping \| passed \| failed \| cancelled`.
`list_runs`	`{ workspace_id?, test_id?, status?, engine?, q?, from?, to?, page?, response_format? }`	Paginated run list.
`cancel_run` (write, destructive)	`{ run_id, mode?: 'graceful'\|'kill', reason? }`	Cancels a run — cannot be undone. Not present in the tool catalog at all under a read-only session.
`rerun_run` (write)	`{ run_id, source: 'run_snapshot'\|'current_test' }`	New run derived from an existing one.
`add_runners` (write)	`{ run_id, additions: [{ location_plan_row_id, count }] }`	Scales up a live run at existing location-plan rows.

Results and diagnostics

Tool	Args	Returns
`get_run_results`	`{ run_id, ...optional time-range/location/label/runner/scenario filters }`	KPI summary (throughput, latency, error rate, duration).
`query_run_metrics`	`{ run_id, queries: [{ category, aggregation, alignment?, filters?, group_by? }] }` (≤8 queries, ≤5 filters/query, ≤3 group-bys/query)	Time series for the requested metrics.
`get_run_errors`	`{ run_id, response_format? }`	Grouped error rows (message/count/code).
`get_run_summary`	`{ run_id, transactions?, response_format? }`	Executive summary: headline outcome + tripped failure criteria.
`get_run_error_bodies`	`{ run_id, row_id?, from?, to?, runner?, label?, scenario?, error_key?, limit?, response_format? }`	Sampled error request/response bodies + status codes.
`get_run_logs`	`{ run_id, severity?: 'error'\|'warn', stream?: 'all'\|'system'\|'engine', from?, to?, limit?, cursor?, response_format? }`	Error-level engine/system log lines.
`get_runner_health`	`{ run_id, from?, to?, runners?, response_format? }`	Runner heartbeat/CPU/mem trend + VU shortfall.
`detect_run_anomalies`	`{ run_id, metric_ids?, sensitivity?: 'low'\|'normal'\|'high', allow_partial? }`	Robust outlier detection on the run’s charts (terminal runs only).

Dashboards

Tool	Args	Returns
`get_metrics_catalog`	(none)	The set of chartable metrics.
`list_dashboards`	(none)	Dashboards visible in the account.
`get_dashboard`	`{ dashboard_id }`	One dashboard’s panels/config.
`create_dashboard` (write)	`{ title, ... }`	New dashboard.
`update_dashboard` (write)	`{ dashboard_id, title?, ... }`	Updated dashboard.

Browser fleets

Managed Selenium-Grid / Playwright-connect() capacity — N browsers behind one endpoint, driven directly by your own client (not a MaxoPerf-orchestrated test run). See Browser fleets.

Tool	Args	Returns
`create_browser_fleet` (write, idempotent)	`{ browsers: [{ browser: 'chrome'\|'firefox'\|'edge', count, location }], browsers_per_runner?, ttl_minutes?, idle_timeout_minutes?, name? }` — `location` is `<aws\|gcp\|azure>:<region>` or `byoc:<pdcId>`	`{ fleetId, runId, status, endpoints, token, expiresAt, supportedPlaywrightVersions }`. The token is shown only here — re-fetch it with `get_browser_fleet`.
`get_browser_fleet`	`{ fleet_id, include_token? }` (default `true`)	Status, per-location readiness, endpoints, session count, and (unless omitted) the reconnect token.
`list_browser_fleets`	`{ page?, page_size? }`	`{ fleets, total, page, pageSize }` — never includes tokens.
`scale_browser_fleet` (write)	`{ fleet_id, browsers: [...] }`	Updated fleet detail. Scale-down isn’t supported — delete and recreate smaller.
`delete_browser_fleet` (write, destructive)	`{ fleet_id }`	`{ fleetId, terminated: true }`. Idempotent on an already-terminal fleet.

Tunnels

Expose a locally-running or not-yet-deployed target at a public https://tunnel-<name>.maxoperf.com URL. See Tunnels.

Tool	Args	Returns
`create_tunnel` (write)	`{ name, auth_mode?: 'none'\|'basic'\|'bearer', allow_ips?, rate_limit_rpm?, idle_timeout_seconds?, capture_mode?: 'none'\|'headers'\|'body' }` — `name` is a globally-unique DNS label	`{ id, name, url, status, online, clientToken, runCommand, ... }`. The `clientToken` is shown only here.
`list_tunnels`	`{ page?, page_size? }`	Tunnel list — no client tokens.
`get_tunnel`	`{ tunnel_id }`	One tunnel’s detail.
`stop_tunnel` (write)	`{ tunnel_id }`	Disconnects the live client; the URL is preserved but 503s until restarted.
`get_tunnel_stats`	`{ tunnel_id, window_minutes?, step_seconds? }`	Requests, 4xx/5xx, blocked, bytes, p50/p95/p99 latency over a recent window.

Virtual services

A mock/simulated backend deployed as its own pod on a chosen outpost. See Virtual services.

Tool	Args	Returns
`list_virtual_services`	`{ workspace_id }`	`{ virtualServices: [{ id, state, placement, hostname, endpointUrl }] }`
`get_virtual_service`	`{ virtual_service_id }`	Full detail (state, placement, noMatch policy, transaction selection, idle-shutdown config).
`create_virtual_service` (write)	`{ workspace_id, service_id, name, placement?: { private_datacenter_id? }, no_match?, think_scale_percent?, transaction_selection?, idle_shutdown_minutes? }`	New virtual service in draft/stopped state.
`deploy_virtual_service` (write)	`{ virtual_service_id }`	State moving toward `running`. A PDC must be chosen first.
`stop_virtual_service` (write)	`{ virtual_service_id }`	State moving toward `stopped`.
`delete_virtual_service` (write, destructive)	`{ virtual_service_id }`	Deletes the virtual service.

Execution locations and on-demand browser

Tool	Args	Returns
`list_execution_locations`	`{ workspace_id? }`	`{ managed: { providers }, byoc: { privateDatacenters } }` — `byoc.privateDatacenters` merges the shared, MaxoPerf-operated Shared Sandbox outpost with any self-hosted outpost your workspace registered. Use the `id` as `location: "byoc:<id>"` (fleets) or `placement.privateDatacenterId` (virtual services).
`inspect_url` (write)	`{ url }`	Spins up (or reuses) a real browser, navigates to `url`, and returns the JS-rendered DOM/text/outline — use it instead of `fetch_url`/`read_page_dom` when the target is a single-page app.

Escape hatch

For the long tail of Platform API endpoints not covered by a named tool above — billing, secrets, notifications, schedules, and more.

Tool	Args	Returns
`call_platform_api`	`{ method: 'GET'\|'POST'\|'PATCH'\|'DELETE', path, body? }`	Calls any public endpoint. Admin/internal/key-management paths are always rejected before any upstream call. Read-only sessions may only `GET`.
`get_openapi`	(none)	The public OpenAPI document. Large — prefer `search_endpoints` first.
`search_endpoints`	`{ query }`	Keyword search over the OpenAPI document.

Docs and the skill

Tool	Args	Returns
`search_docs`	`{ query }`	Keyword search over this documentation site’s build-time index.
`get_agent_skill`	(varies)	The MaxoPerf agent skill’s instructions — see MaxoPerf agents.
`list_agent_skill_sections`	(none)	The skill bundle’s section index.

Web research

All reads, all funneled through one SSRF-guarded fetch. web_search self-disables (drops out of the tool catalog) when no search provider is configured for the deployment.

Tool	Args	Returns
`web_search`	(query args)	Web search results, when a provider is configured.
`fetch_url`	`{ url, ... }`	Static page fetch — HTML only, no JavaScript execution.
`read_openapi`	`{ url, ... }`	Fetches and parses a third-party OpenAPI document.
`read_page_dom`	`{ url, ... }`	Static DOM extraction — same JS limitation as `fetch_url`.

Prompts

Prompts are pre-built, multi-step recipes a client can surface as one-click flows (prompts/list) instead of an agent rediscovering the right tool-call sequence itself.

Prompt	Args	What it guides
`run-baseline-load-test`	`{ test_id }`	Start a run and report pass/fail, duration, throughput, p95 latency, and error rate once it’s terminal.
`diagnose-latency-regression`	`{ baseline_run_id, candidate_run_id }`	Compare p95 latency between two runs of the same test and correlate with errors.
`summarize-run`	`{ run_id }`	A concise, plain-language summary of one run.
`plan-and-build-test`	`{ goal, target? }`	Turn a plain-language goal into the project → test → upload → run sequence.
`choose-executor`	`{ test_kind }`	Recommend a test engine/executor with rationale.
`scan-endpoints-for-hotspots`	`{ openapi_or_paste }`	Rank likely load-testing hotspots from an OpenAPI spec or pasted endpoint list.
`setup-secrets-and-envs`	`{ test_id }`	Guide configuring secrets and multi-environment values before the first run.
`diagnose-run-failure`	`{ run_id }`	Root-cause a failed/underperforming run: tripped criteria, error bodies, logs, runner health, anomalies.
`explain-run-anomalies`	`{ run_id }`	Run the anomaly scan on a finished run and explain flagged outliers.

Resources

Two read-only MCP resources for clients that prefer resources over tool calls:

Resource URI	Contents
`maxoperf://openapi`	The public Platform API OpenAPI document (same source as `get_openapi`).
`maxoperf://run/{id}`	A run’s current status + dashboard overview.

Error envelope

Every tool error follows the same shape: { isError: true, content: [{ type: "text", text: "<status> <code>: <detail>" }], structuredContent: { status, code, detail } }. Common codes: READ_ONLY_MODE, FORBIDDEN_PATH (call_platform_api deny-listed path), TOO_MANY_QUERIES / TOO_MANY_FILTERS / TOO_MANY_SERIES (query_run_metrics limits), UPLOAD_FAILED, PATH_NOT_ALLOWED (upload_test_file local-path traversal), INVALID_ARGUMENTS.