# AX Check: mindshub.ai Checked 2026-10-07. Docs and pricing are clear, but one agent couldn't pin down cost. 16 of 23 checks passed: quickstart, code samples and auth steps are all usable. Pricing is published openly, yet 1 of 3 agent sessions still said it couldn't state a price. ## Onboarding needs a login ## Coding sessions All three sessions completed. DeepSeek V4.1 Flash and Qwen 3.8 Max both reported free-tier and prepaid credit pricing with per-model token rates pulled from the live rate card. Kimi K3 completed its run but stated it could not determine pricing and declined to guess a figure. ### DeepSeek V4.1 Flash [View public run](https://agents.withgauge.com/p/runs/de07ab61-d0d4-42ba-8747-410b8ae6c39d) · [Read transcript](https://www.ax-check.com/mindshub.ai/sessions/deepseek.json) README.md and final summary state free-tier vs prepaid credit pricing with per-token rates by model, noting these are drawn from the live rate card at mindshub.ai/pricing and may change. #### End-to-end onboarding - **Onboarding needs a login**: The agent never obtained real MindsHub credentials. It searched the environment for an existing API key (none found), then confirmed account/key creation requires a human to sign up at the console. All authenticated calls used a fabricated dummy key and correctly received 401s, proving no self-service credential acquisition occurred during the session. Event 58: ```text no MINDSHUB_API_KEY in env ``` Event 80: ```text HTTP 401 {"type":"error","error":{"type":"authentication_error","message":"Invalid or missing credentials. See https://docs.mindshub.ai/inference/authentication"}} ``` Event 94: ```text I can't execute a **live inference call**, because that needs a `MINDSHUB_API_KEY`, and key creation requires a human to sign up at `https://console.mindshub.ai`. ``` #### Hallucinated URLs None identified in this transcript. #### Blockers - **API key creation requires human sign-up at the console**: MindsHub Inference requires an API key obtained by signing up at console.mindshub.ai; there is no programmatic or anonymous key-issuance path the agent could use. This is normal product authentication behavior, not a defect, but it capped the session at building an unauthenticated example rather than exercising a live call. Event 94: ```text key creation requires a human to sign up at `https://console.mindshub.ai` ``` Event 57: ```text Invalid or missing credentials. See https://docs.mindshub.ai/inference/authentication ``` #### Suggested Changes - **Document the Cloudflare User-Agent requirement near authentication docs**: A stdlib urllib request with the default User-Agent was rejected with Cloudflare error 1010 instead of a normal auth error. Add a note on the inference getting-started or authentication docs page stating a custom User-Agent header is required, so bare HTTP client users do not confuse an edge block with an API problem. Verify by POSTing to /v1/chat/completions with the default urllib UA and checking the response is a standard auth error, not code 1010. Event 75: ```text HTTP 403 from MindsHub: error code: 1010 ``` Event 82: ```text the default `Python-urllib` User-Agent was blocked by Cloudflare. Adding a User-Agent fixes it. ``` ### Kimi K3 [View public run](https://agents.withgauge.com/p/runs/84f8dc08-e5a3-4f33-b521-3dff8d09d51f) · [Read transcript](https://www.ax-check.com/mindshub.ai/sessions/kimi.json) Final output explicitly states 'I can't tell you' on pricing and refuses to invent a number — no cost figure or assumptions are given at all. #### End-to-end onboarding - **End-to-end onboarding not demonstrated**: The agent never obtained credentials or reached a real product. It searched PyPI, npm, the product's domain, and a web search, found no package, a 502 on the website, and zero search results, then stopped without attempting any authenticated operation. Event 14: ```text I can't find any product called "MindsHub" — no hosted service, SDK, API, or documentation exists under that name that I can locate. ``` #### Hallucinated URLs - **Guessed root domain for MindsHub**: The agent guessed https://mindshub.com as the product's website without any documentation pointing to it. The request returned a 502, which alone does not prove the URL was invented (it could be a real but down domain), so this is reported as an attempted guess rather than a confirmed hallucination. Event 5: ```text https://mindshub.com ``` Event 7: ```text 502 ``` #### Blockers - **Product MindsHub could not be located anywhere**: No package registry (PyPI, npm), no working website, and no search engine results surfaced any trace of a product called MindsHub. This is not a credentials problem or agent error — the agent ran reasonable discovery checks across standard channels and found nothing to build against, so it correctly stopped per the user's instruction rather than inventing an SDK or API. Event 7: ```text ERROR: No matching distribution found for mindshub ``` Event 7: ```text npm error 404 'mindshub@*' is not in this registry. ``` Event 7: ```text 502 ``` Event 12: ```text "RelatedTopics":[],"Results":[] ``` #### Suggested Changes None identified in this transcript. ### Qwen 3.8 Max [View public run](https://agents.withgauge.com/p/runs/c493aba0-7927-4986-81b2-e16d8fb5f7e6) · [Read transcript](https://www.ax-check.com/mindshub.ai/sessions/qwen.json) README.md pricing section and final summary cite wallet model, $10 min top-up, free-tier model (mindshub_air), per-model per-1M-token rates with a dated snapshot ('Sept 2026'), and caching discount — explicit assumptions behind the figures, sourced from real curl fetches of mindshub.ai/pricing (seq 28-37). #### End-to-end onboarding - **Onboarding needs a login**: The agent never obtained a real MindsHub API key. It searched the sandbox environment for credentials (none found), then tested its example scripts against a self-supplied placeholder key (mdb_test_invalid) and got 401 authentication_error from the live API both times. The agent explicitly states signup requires a browser, email, and reCAPTCHA that it cannot complete autonomously, and stops there, recommending the human export a real key. Event 45: ```text 401 {"type":"error","error":{"type":"authentication_error","message":"Invalid or missing credentials. See https://docs.mindshub.ai/inference/authentication"}} ``` Event 86: ```text openai.AuthenticationError: Error code: 401 - {'type': 'error', 'error': {'type': 'authentication_error', 'message': 'Invalid or missing credentials. See https://docs.mindshub.ai/inference/authentication'}} ``` Event 100: ```text there's no `MINDSHUB_API_KEY` in this environment, and getting one requires signing up at console.mindshub.ai (browser + email + reCAPTCHA), which I can't do for you. ``` #### Hallucinated URLs None identified in this transcript. #### Blockers - **No self-service way to obtain a MindsHub API key**: Product behavior / session limitation, not an agent error: MindsHub requires signup through a web console (console.mindshub.ai) to mint an API key, and the sandbox had no pre-provisioned credential. The agent correctly identified this and tried a placeholder key, confirming the live API rejects unauthenticated/invalid requests with a clean 401, but could not proceed to an authenticated call. Event 46: ```text 401 {"type":"error","error":{"type":"authentication_error","message":"Invalid or missing credentials. See https://docs.mindshub.ai/inference/authentication"}} ``` Event 100: ```text getting one requires signing up at console.mindshub.ai (browser + email + reCAPTCHA), which I can't do for you ``` - **Decisions endpoint silently rejects a bare HTTP client's default user agent**: Test-environment/product edge case: calling POST /v1/decisions with Python's urllib default headers returned a Cloudflare-level 403 (code 1010) instead of the API's normal 401 for a bad key, momentarily masking the real auth error. The agent diagnosed this itself and added an explicit User-Agent header, after which the response correctly surfaced as 401 authentication_error. Event 86: ```text HTTP 403: error code: 1010 ``` Event 98: ```text HTTP 401: {"type":"error","error":{"type":"authentication_error","message":"Invalid or missing credentials. See https://docs.mindshub.ai/inference/authentication"}} ``` #### Suggested Changes - **Document the urllib/Cloudflare user-agent 403 on the Decisions page**: On docs.mindshub.ai/inference/decisions, add a note that requests from clients with no or a minimal User-Agent (e.g. Python's urllib default) can be blocked by Cloudflare with a 403 code 1010 before reaching MindsHub's own auth check, and that setting a custom User-Agent header resolves it. Verify by re-running the stdlib urllib sample from the docs without a custom header and confirming the 403 reproduces, then confirming it clears once a User-Agent is added. Event 86: ```text HTTP 403: error code: 1010 ``` Event 94: ```text Cloudflare rejects urllib's default user agent (403, code 1010). ``` ### Task given to each agent Help me build a simple example using MindsHub. 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: B · 84/100 (provisional) Grades come from completed site checks. Coding sessions and skipped checks do not affect the score. ### Clarity - **Failed** — Homepage answers Markdown requests ```text Homepage returned text/html for a text/markdown request; no Markdown representation offered. ``` - **Pass** — llms.txt provides an actionable documentation index ```text llms.txt lists core pages, docs, pricing, models, blog and legal with descriptive links. ``` - **Pass** — llms.txt provides navigation guidance ```text llms.txt groups links under Core pages, Blog, Newsroom, Events and Legal headings. ``` - **Pass** — llms.txt mentions offered API, MCP, and skills ```text llms.txt describes MindsHub Inference API, Jev Decisions API and SDK-compatible endpoints. ``` - **Pass** — A compact guide representation exists ```text Site publishes standalone Markdown guides, e.g. getting-started.md and choosing-an-api.md, served as text/markdown. ``` - **Pass** — A focused guide is directly retrievable ```text getting-started.md is a focused, directly retrievable guide with key setup and runnable samples. ``` - **Pass** — Equivalent instructions fit a token budget ```text getting-started.md is 2318 tokens, well under the 8000-token budget, and the workflow stays usable. ``` - **Skipped** — Product-docs links survive format changes ```text Homepage Markdown is unsupported, so link preservation across format changes cannot be measured. ``` - **Pass** — The compact guide is independently actionable ```text Getting-started guide gives concrete steps: sign up, create key, run Python/TS samples. ``` - **Pass** — Install and next-step links resolve ```text Fetched install/next-step docs (getting-started, choosing-an-api) returned HTTP 200. ``` ### Onboarding - **Pass** — Docs lead to a relevant quickstart ```text Getting-started page gives concrete first steps: sign up, create key, run sample requests. ``` - **Pass** — Installation commands are extractable ```text Install/usage commands extractable: pip openai/anthropic SDKs, export MINDSHUB_API_KEY, curl usage call. ``` - **Pass** — Code examples are available without interaction ```text Python and TypeScript code samples shown inline in tabs, no interaction required. ``` - **Pass** — Prerequisites and auth boundaries are explicit ```text Auth explicit: sign up at console.mindshub.ai, create API key, set MINDSHUB_API_KEY. ``` ### Pricing - **Failed** — Agents identify pricing and its assumptions ```text 3 of 3 sessions were judged on pricing; 1 fell short. DeepSeek V4.1 Flash: README.md and final summary state free-tier vs prepaid credit pricing with per-token rates by model, noting these are drawn from the live rate card at mindshub.ai/pricing and may change. Kimi K3: Final output explicitly states 'I can't tell you' on pricing and refuses to invent a number — no cost figure or assumptions are given at all. Qwen 3.8 Max: README.md pricing section and final summary cite wallet model, $10 min top-up, free-tier model (mindshub_air), per-model per-1M-token rates with a dated snapshot ('Sept 2026'), and caching discount — explicit assumptions behind the figures, sourced from real curl fetches of mindshub.ai/pricing (seq 28-37). This behavioural item does not affect the fast grade. ``` - **Pass** — Pricing is readable without interaction ```text Pricing page renders full rate card and FAQ as static text, no interaction needed. ``` - **Pass** — Prices are stated, not gated ```text Free tier $0 and per-model token rates published openly; no gating. ``` - **Pass** — Pricing units and limits are explicit ```text Units explicit: USD per 1M tokens, per 1,000 searches; daily fair-use limits noted. ``` ### Activation - **Pass** — An API reference or OpenAPI spec is reachable ```text API reference with endpoints (chat, responses, messages, decisions) and bearer auth is reachable. ``` - **Skipped** — An MCP server is documented and well-formed ```text No MCP server documentation found in the fetched evidence. ``` - **Skipped** — A CLI install path is documented ```text No CLI install path documented; docs cover SDKs and coding agents, not a CLI. ``` - **Skipped** — SDK packages resolve on their registries ```text No registry lookup for a MindsHub SDK package was supplied. ``` - **Skipped** — Agent skills are published ```text No agent skills published in the fetched evidence. ``` [Full report data](https://www.ax-check.com/mindshub.ai/report.json)