# AX Check: testmuai.com
Checked 2026-10-15.

Agents can install, price, and ship against this product fast
All 23 checklist items passed: docs, CLI, API, MCP, and pricing are all machine-readable with no login wall. 3 of 3 sessions found clear USD pricing.

## Onboarding needs a login

## Coding sessions
Three independent sessions (DeepSeek V4.1 Flash, Kimi K3, Qwen 3.8 Max) each read the pricing page directly and reported concrete USD prices with billing units, credit quotas, and free-tier terms, with DeepSeek also flagging a docs conflict on Kane CLI free credits. All three completed without hitting a login wall.

### DeepSeek V4.1 Flash
[View public run](https://agents.withgauge.com/p/runs/288ff071-a56b-4f0a-b1b4-925873de161b) · [Read transcript](https://www.ax-check.com/testmuai.com/sessions/deepseek.json)
Quoted USD prices by product, with the billing unit, annual vs. monthly rates, credit quotas, free tiers and trials. It also pointed out a conflict between the two docs on Kane CLI free credits. (seq 24, seq 23, seq 67)
#### End-to-end onboarding
- **Onboarding needs a login**: The agent researched TestMu AI's docs thoroughly and built a working example script, but never obtained real account credentials. It explicitly states the sandbox has no TestMu AI account and it cannot sign up on the user's behalf. It only validated the script's failure paths (missing creds, bogus creds against the live API returning 401) — never a successful authenticated operation.
  Event 67:

  ```text
  **No credentials.** This sandbox has no TestMu AI account, username, or access key, and I can't sign up on your behalf.
  ```
  Event 65:

  ```text
  === bogus creds (expect live 401, exit 2) ===
  Request failed: HTTP 401 Unauthorized
  HTTP Basic: Access denied.
  exit=2
  
  ```
  Event 67:

  ```text
  Everything else is ready; drop in `LT_USERNAME` / `LT_ACCESS_KEY` and the example runs.
  ```

#### Hallucinated URLs
None identified in this transcript.

#### Blockers
- **No account credentials available for the hosted API or CLI**: The sandbox had no TestMu AI username or access key, and the agent has no self-service way to create an account. This blocked any authenticated call to the Automation REST API or Kane CLI/MCP flows beyond testing the 401 error path. This is a session/environment limitation (missing human-supplied credentials), not a product defect — TestMu AI requires a normal signup/login like most SaaS products.
  Event 67:

  ```text
  **No credentials.** This sandbox has no TestMu AI account, username, or access key, and I can't sign up on your behalf. That's the one thing preventing the happy-path run and any real browser session.
  ```
- **MCP server setup requires interactive browser OAuth**: The MCP server's auth model opens a browser for consent on testmuai.com, which cannot be completed in a headless agent session. This is expected product behavior (OAuth is the documented auth method for MCP) rather than a bug, but it did stop the agent from exercising that surface.
  Event 67:

  ```text
  **MCP OAuth is interactive.** Setting up the MCP server requires a browser consent on testmuai.com, which I can't complete headlessly here.
  ```

#### Suggested Changes
- **Reconcile conflicting Kane CLI free-tier claims across docs**: agents.md states Kane CLI has a free tier of 200 credits/month, while pricing.md states the Kane CLI is free with no separate paid plan and that KaneAI (not Kane CLI) has no free plan. Update one of these two machine-readable docs (https://www.testmuai.com/agents.md and https://www.testmuai.com/pricing.md) so the credit/free-tier language matches, and verify by re-fetching both pricing.md and agents.md and checking the Kane CLI free-tier description agrees.
  Event 17:

  ```text
  | Kane CLI (browser automation from an agent) | Yes, 200 credits/month | Starter $19/mo, Pro $99/mo, Enterprise custom |
  ```
  Event 24:

  ```text
  The CLI itself is free to install and use.
  - Local runs are free. Local authoring through Kane CLI is included in every KaneAI credit plan (it is the only authoring mode on Starter).
  ```

### Kimi K3
[View public run](https://agents.withgauge.com/p/runs/ea5c4eef-9423-4d8f-b537-64cdf25656da) · [Read transcript](https://www.ax-check.com/testmuai.com/sessions/kimi.json)
Gave per-product prices from the fetched pricing.md. Named the plan tier, the unit (per parallel, agent, or user), annual vs. monthly billing, credit use, free tiers and non-cumulative tracks. (seq 24, seq 100)
#### End-to-end onboarding
- **Onboarding needs a login**: The session never obtained real TestMu AI credentials. No account/API key existed in the sandbox, OAuth login requires an interactive browser unavailable here, and the agent explicitly flagged this as a blocker rather than working around it. All example scripts were written to require a human-supplied TESTMUAI_USERNAME/TESTMUAI_ACCESS_KEY and were never actually executed against the hosted API or CLI — no authenticated operation occurred.
  Event 100:

  ```text
  No credentials: no TestMu AI account/key in this environment, and OAuth login needs an interactive browser — so I couldn't execute a live run. With `TESTMUAI_USERNAME`/`TESTMUAI_ACCESS_KEY` set, both scripts should work
  ```
  Event 96:

  ```text
  Successfully wrote 1897 bytes to /sandbox/repo/README.md
  ```
  Event 4:

  ```text
  env | grep -iE 'testmu|lambdatest|LT_'
  ```

#### Hallucinated URLs
None identified in this transcript.

#### Blockers
- **No TestMu AI account credentials available**: The sandbox had no TestMu AI/LambdaTest username or access key, and interactive OAuth login was impossible in this non-interactive environment. This is a test-environment limitation (missing credentials), not a product defect — basic auth via CI flags is documented as the intended workaround, but no key was ever provisioned to use it.
  Event 4:

  ```text
  env | grep -iE 'testmu|lambdatest|LT_' ; echo "---"; node --version 2>/dev/null; python3 --version 2>/dev/null; echo "---"; ls /sandbox/repo
  ```
  Event 100:

  ```text
  No credentials: no TestMu AI account/key in this environment, and OAuth login needs an interactive browser — so I couldn't execute a live run.
  ```
- **Kane CLI global install repeatedly timed out**: Agent error / environment constraint: `npm install -g @testmuai/kane-cli` exceeded both a 120s and a 90s timeout twice due to a heavy dependency tree (React/Ink TUI, pdfjs, etc.), and the resulting partial install left no working `kane-cli` binary on PATH, so the CLI was never actually run.
  Event 60:

  ```text
  Command timed out after 120 seconds
  ```
  Event 81:

  ```text
  Command timed out after 90 seconds
  ```
  Event 72:

  ```text
  /bin/bash: line 1: kane-cli: command not found
  ```
- **No Chrome binary present for local Kane CLI runs**: Product behavior: Kane CLI requires a locally installed Google Chrome (via DevTools Protocol) for local runs, and none was present in the sandbox, which would block any local CLI execution even if install had succeeded.
  Event 53:

  ```text
  ls: cannot access '/usr/bin/google-chrome': No such file or directory
  ls: cannot access '/usr/bin/chromium*': No such file or directory
  ls: cannot access '/opt/google': No such file or directory
  NO CHROME
  ```

#### Suggested Changes
- **Reduce or document the Kane CLI's install footprint**: In the Kane CLI README's Install section (raw.githubusercontent.com/LambdaTest/kane-cli/main/README.md), add an expected install-time estimate or a lighter-weight install path, since `npm install -g @testmuai/kane-cli` exceeded 120 seconds twice in a fresh sandbox due to its dependency tree (React/Ink TUI, pdfjs-dist, aws4fetch, etc.). Verify by timing a clean `npm install -g @testmuai/kane-cli` on a fresh container and confirming it completes well under common CI timeout defaults (e.g., 60-90s).
  Event 60:

  ```text
  Command timed out after 120 seconds
  ```
  Event 32:

  ```text
  @testmuai/kane-cli@0.8.23 | Apache-2.0 | deps: 17 | versions: 68
  ```

### Qwen 3.8 Max
[View public run](https://agents.withgauge.com/p/runs/ddebf9ab-adb5-4df9-b22e-aee2a24bec7e) · [Read transcript](https://www.ax-check.com/testmuai.com/sessions/qwen.json)
Quoted KaneAI Starter $17/mo and Pro $89/mo from the pricing page, with assumptions: billed annually, per agent, 2,000 vs 12,000 credits, free tier. Browser Cloud cost was not given. (seq 34, seq 82)
#### End-to-end onboarding
- **Onboarding needs a login**: Agent built a working local example calling the hosted TestMu AI Browser Cloud SDK, but had no LT_USERNAME/LT_ACCESS_KEY credentials available in the sandbox and explicitly declined to self-register an account. The script was run and failed fast on the missing-credentials check; no authenticated call to the real product (session creation) was ever made.
  Event 6:

  ```text
  Command exited with code 2
  ```
  Event 75:

  ```text
  Missing credentials: set LT_USERNAME and LT_ACCESS_KEY (see README).
  
  
  Command exited with code 1
  ```
  Event 82:

  ```text
  I'm blocked at the last step: **no TestMu AI credentials exist in this environment** (`LT_USERNAME` / `LT_ACCESS_KEY` are unset, and I won't create an account on your behalf).
  ```

#### Hallucinated URLs
None identified in this transcript.

#### Blockers
- **No TestMu AI account credentials available**: The sandbox had no LT_USERNAME/LT_ACCESS_KEY set, and the agent correctly chose not to self-register an account on the user's behalf. This is a normal login/credentials requirement, not a product defect — the quickstart clearly documents how to obtain credentials via the dashboard.
  Event 6:

  ```text
  Command exited with code 2
  ```
  Event 75:

  ```text
  Missing credentials: set LT_USERNAME and LT_ACCESS_KEY (see README).
  ```
  Event 82:

  ```text
  no TestMu AI credentials exist in this environment
  ```
- **Legacy Screenshot REST API returned 502 and api.lambdatest.com returned 403**: Early probes of the general API host and the older Automated Screenshot REST endpoint failed outright (403 on the bare domain, 502 on the specific os-browsers endpoint) even without credentials, which muddied discovery before the agent found the correct current path (the Node SDK for Browser Cloud). This looks like either edge/gateway behavior on LambdaTest's infrastructure or an environment network restriction rather than something the agent caused.
  Event 12:

  ```text
  403
  {}
  ERROR: No matching distribution found for testmu
  ```
  Event 42:

  ```text
  error code: 502
  
  HTTP 502
  ```
- **Browser Cloud REST API not yet available**: The documented REST API for Browser Cloud is explicitly marked as not yet shipped, forcing reliance on the Node SDK only. This is disclosed product status, not an agent error, but it limits workflow options for non-Node stacks.
  Event 49:

  ```text
  # API Reference (Coming Soon)
  
  > For the full site index for AI agents, see [llms.txt](https://www.testmuai.com/support/docs/llms.txt).
  
  The REST API will expose all TestMu AI Browser SDK capabilities over HTTP, making
  TestMu AI Browser Cloud accessible from any programming language.
  ```

#### Suggested Changes
- **Fix or redirect the broken Automated Screenshot API host**: Requests to https://api.lambdatest.com (bare) and https://api.lambdatest.com/screenshots/v1/os-browsers returned 403 and 502 respectively during this session. Check the gateway/load balancer config for that legacy API host so documented example curl calls in automated-screenshot-api-for-cross-browser-testing.md actually resolve, and verify by re-running the exact curl command from the doc.
  Event 12:

  ```text
  403
  {}
  ERROR: No matching distribution found for testmu
  ```
  Event 42:

  ```text
  error code: 502
  
  HTTP 502
  ```
- **Ship the Browser Cloud REST API or clearly flag SDK-only status earlier in docs**: The API Reference page for Browser Cloud is a placeholder ('Coming Soon') discovered only after the agent had already explored several other doc pages. Surface this SDK-only limitation directly in the llms.txt summary line or at the top of browser-cloud-api-reference.md so agents and developers evaluating non-Node integration paths see it immediately instead of after multiple fetches.
  Event 49:

  ```text
  # API Reference (Coming Soon)
  
  > For the full site index for AI agents, see [llms.txt](https://www.testmuai.com/support/docs/llms.txt).
  
  The REST API will expose all TestMu AI Browser SDK capabilities over HTTP
  ```

### Task given to each agent
Help me build a simple example using TestMu AI. Tell me how pricing works, and briefly tell me whether this product will be easy for you to manage. Let me know if you get blocked. If this product has no developer workflow you can act on, say so plainly and stop. Stay light: use the hosted product through its SDK or API. Do not start local service stacks or wait for long-running commands; if the quickstart requires either, say so plainly and stop.

No product credentials were supplied and no purchases were authorized.

## Score: A · 100/100 (provisional)
Grades come from completed site checks. Coding sessions and skipped checks do not affect the score.

### Clarity
- **Pass** — Homepage answers Markdown requests

  ```text
  Homepage returned text/markdown (200) when requested with Accept: text/markdown.
  ```

- **Pass** — llms.txt provides an actionable documentation index

  ```text
  llms.txt (200) links docs, API, OpenAPI, MCP, CLI, skills with organized sections.
  ```

- **Pass** — llms.txt provides navigation guidance

  ```text
  llms.txt groups surfaces with guidance for AI assistants and coding agents.
  ```

- **Pass** — llms.txt mentions offered API, MCP, and skills

  ```text
  llms.txt details MCP server URL, Kane CLI install, and agent skill files.
  ```

- **Pass** — A compact guide representation exists

  ```text
  llms.txt links agent-readable Markdown guides: agents.md, SKILL.md, kane-cli/agents.md, pricing.md.
  ```

- **Pass** — A focused guide is directly retrievable

  ```text
  kane-cli/agents.md is a standalone Markdown guide fetched directly with concrete first-run steps.
  ```

- **Pass** — Equivalent instructions fit a token budget

  ```text
  kane-cli/agents.md is 1903 tokens, well under the 8000-token budget, and keeps the workflow usable.
  ```

- **Pass** — Product-docs links survive format changes

  ```text
  Homepage Markdown is supported and retains product-docs links like /support/docs/ and /mcp/.
  ```

- **Pass** — The compact guide is independently actionable

  ```text
  Kane CLI agents.md gives install, first-run, preferences, CI steps; SKILL.md gives full MCP setup.
  ```

- **Pass** — Install and next-step links resolve

  ```text
  Fetched install/next-step pages (launch-first-session, KaneAI docs, MCP docs) all returned 200.
  ```


### Onboarding
- **Pass** — Docs lead to a relevant quickstart

  ```text
  Docs index links quickstarts: Browser Cloud launch-first-session and Kane CLI quickstart.
  ```

- **Pass** — Installation commands are extractable

  ```text
  Install commands extractable: npm i @testmuai/browser-cloud and npm install -g @testmuai/kane-cli.
  ```

- **Pass** — Code examples are available without interaction

  ```text
  Full TypeScript Puppeteer/Playwright/Selenium examples shown inline without interaction.
  ```

- **Pass** — Prerequisites and auth boundaries are explicit

  ```text
  Prerequisites list signup, Username/Access Key from Settings, and .env setup.
  ```


### Pricing
- **Pass** — Pricing is readable without interaction

  ```text
  Pricing page renders full plan tables and prices as static text, no interaction needed.
  ```

- **Pass** — Prices are stated, not gated

  ```text
  Concrete USD prices listed per plan, e.g. KaneAI $19/agent/mo, Automation $29/mo annual.
  ```

- **Pass** — Pricing units and limits are explicit

  ```text
  Units explicit: per agent credits, per parallel session, per user, VUH, per screenshot.
  ```

- **Pass** — Agents identify pricing and its assumptions

  ```text
  3 of 3 sessions were judged on pricing; 0 fell short. DeepSeek V4.1 Flash: Quoted USD prices by product, with the billing unit, annual vs. monthly rates, credit quotas, free tiers and trials. It also pointed out a conflict between the two docs on Kane CLI free credits. (seq 24, seq 23, seq 67) Kimi K3: Gave per-product prices from the fetched pricing.md. Named the plan tier, the unit (per parallel, agent, or user), annual vs. monthly billing, credit use, free tiers and non-cumulative tracks. (seq 24, seq 100) Qwen 3.8 Max: Quoted KaneAI Starter $17/mo and Pro $89/mo from the pricing page, with assumptions: billed annually, per agent, 2,000 vs 12,000 credits, free tier. Browser Cloud cost was not given. (seq 34, seq 82) This behavioural item does not affect the fast grade.
  ```


### Activation
- **Pass** — An API reference or OpenAPI spec is reachable

  ```text
  OpenAPI.md serves combined OpenAPI/Swagger specs for 12 TestMu AI REST APIs; api.lambdatest.com responds.
  ```

- **Pass** — An MCP server is documented and well-formed

  ```text
  MCP server documented: URL https://mcp.lambdatest.com/mcp, OAuth, Streamable HTTP/STDIO, five tool groups, client configs.
  ```

- **Pass** — A CLI install path is documented

  ```text
  Kane CLI install documented: npm install -g @testmuai/kane-cli, plus Homebrew and shell installer.
  ```

- **Pass** — SDK packages resolve on their registries

  ```text
  npm registry lookups for @testmuai/kane-cli, @testmuai/browser-cloud, @lambdatest/smartui-cli all returned HTTP 200.
  ```

- **Pass** — Agent skills are published

  ```text
  Agent skills published: SKILL.md, kane-cli/agents.md, kane-ai/agents.md, agent-testing/agents.md, plus agent-skills repo.
  ```



[Full report data](https://www.ax-check.com/testmuai.com/report.json)