# AX Check: lightsage.com Checked 2026-11-11. Docs and pricing are clear, but no setup path was ever tested. 15 of 23 checks passed, all in clarity and pricing. 8 checks on quickstart, install, API and MCP docs went unassessed — no agent reached them. ## Onboarding needs a login ## Coding sessions All three sessions completed. Two, DeepSeek V4.1 Flash and Qwen 3.8 Max, read the pricing page and quoted matching Free/Scale/Enterprise tiers with credit costs. Kimi K3 could not find the product at all and refused to guess at pricing rather than invent a figure. ### DeepSeek V4.1 Flash [View public run](https://agents.withgauge.com/p/runs/af9a256f-74c5-4feb-8d7b-7014fc59651c) · [Read transcript](https://www.ax-check.com/lightsage.com/sessions/deepseek.json) Quoted the Free ($0), Scale ($1,000/mo for 19,000 credits at $0.0526 each) and Enterprise tiers from the pricing page, with trial limits, top-ups and estimated capacity per workload. (seq 60, seq 141) #### End-to-end onboarding - **Onboarding needs a login**: The agent only ever tested the hosted API with no key and with a self-invented dummy key (ls_cli_dummy), both of which are expected to fail or be no-ops. It never signed up for a Lightsage account, never obtained a real API key, and therefore never completed a single authenticated operation. The agent explicitly states the session is blocked on obtaining a real API key tied to an organization, which requires signing up via the dashboard outside the agent's reach. Event 96: ```text == auth status dummy key HTTP/2 401 content-length: 28 content-type: application/json date: Wed, 11 Nov 2026 05:09:26 GMT server: uvicorn vary: Origin {"detail":"Invalid API key"} ``` Event 139: ```text ### Dummy key (expected: clean 401 error) API status: ok Error: 401 Unauthorized on /v2/auth/status: Invalid API key ``` Event 141: ```text I'm blocked on exactly one thing: a **Lightsage API key tied to an organization that has Visibility or API Performance data**. Creating one requires signing up / OAuth into the dashboard, which I can't do on your behalf. ``` #### Hallucinated URLs - **Guessed docs subdomain that does not resolve**: The agent probed https://docs.lightsage.ai as a guessed documentation URL before confirming the real domain was lightsage.com. The guess failed outright (DNS resolution failure), and the agent moved on to the correct, documented domain rather than relying on the guess, so it was a transient dead-end rather than a URL used downstream. Event 13: ```text == https://docs.lightsage.ai curl: (6) Could not resolve host: docs.lightsage.ai code=000 url=https://docs.lightsage.ai/ ``` #### Blockers - **No self-service path to a real API key**: Getting a working Lightsage API key requires creating an account and organization through the hosted dashboard (Settings > API Keys), which needs human sign-up/OAuth. This is a normal authentication requirement, not a product defect, but it stopped the agent from completing any authenticated operation in this session. Event 141: ```text I'm blocked on exactly one thing: a **Lightsage API key tied to an organization that has Visibility or API Performance data**. Creating one requires signing up / OAuth into the dashboard, which I can't do on your behalf. ``` - **Primary domain is a parked for-sale page, not the product**: The first domain the agent tried, lightsage.ai, resolved to a Spaceship.com domain marketplace listing rather than the actual Lightsage product, which could have derailed a less careful agent. This is an external DNS/domain ownership issue, not a Lightsage-hosted-product defect, and the agent correctly pivoted to lightsage.com. Event 20: ```text Lightsage.ai for sale | Spaceship.com ``` Event 20: ```text domainUrl: 'https://lightsage.ai', domainName: 'Lightsage.ai', ``` #### Suggested Changes - **Link lightsage.ai to the real product or stop advertising it**: Searching and crawling naturally lands on https://lightsage.ai first (alphabetically similar, same brand name), but it is a third-party domain-for-sale page unrelated to the product. Check by curling https://lightsage.ai and confirming it no longer returns a Spaceship.com marketplace listing; redirect it to https://lightsage.com if the org owns it, or ensure no official materials reference the .ai domain. Event 20: ```text Lightsage.ai for sale | Spaceship.com ``` ### Kimi K3 [View public run](https://agents.withgauge.com/p/runs/c89c1a0e-9054-483c-b53f-4568dcf5b0a1) · [Read transcript](https://www.ax-check.com/lightsage.com/sessions/kimi.json) Gave no price. It said plainly that it couldn't verify pricing for a product it couldn't find, and that any figure would be made up. (seq 9) #### End-to-end onboarding - **End-to-end onboarding not demonstrated**: The agent never reached any product-specific onboarding step. It searched npm and PyPI for a package called 'lightsage', found nothing, and concluded the product doesn't exist or isn't discoverable, then stopped per the task's instruction to halt if there is no actionable developer workflow. No credentials were obtained, no API was called, and no authenticated operation was attempted. Event 7: ```text npm error 404 Not Found - GET https://registry.npmjs.org/lightsage - Not found ``` Event 9: ```text I have to stop here and be plain about it: **I can't find any evidence that "Lightsage" is a real product**, so there's no developer workflow I can act on. ``` #### Hallucinated URLs None identified in this transcript. #### Blockers - **Product 'Lightsage' not found in any package registry**: The agent searched npm and PyPI for a package named 'lightsage' and found no matches (404 on npm, no distribution on PyPI). It also checked the local repo for any reference to the product and found none. This is not a product defect or agent error — it reflects that the named product could not be located through the standard public channels the agent checked, so the agent correctly halted rather than fabricate an SDK, pricing, or example. Event 7: ```text ERROR: No matching distribution found for lightsage ``` Event 9: ```text npm registry**: `lightsage` does not exist (404). ``` Event 9: ```text **PyPI**: no package named `lightsage`. ``` #### Suggested Changes None identified in this transcript. ### Qwen 3.8 Max [View public run](https://agents.withgauge.com/p/runs/198955b4-6fa8-4af3-8830-51ffe820908a) · [Read transcript](https://www.ax-check.com/lightsage.com/sessions/qwen.json) Gave Free $0, Scale $1,000/mo with 19,000 credits, and custom Enterprise pricing. It listed per-run credit costs (1 or 5), a usage formula, trial and top-up limits, and flagged outdated pricing in the docs. (seq 23, seq 134, seq 144) #### End-to-end onboarding - **Onboarding needs a login**: Agent explored the Lightsage docs, Public API, Agent Analytics SDK, and OpenAPI spec extensively but never obtained real product credentials. Both the dashboard API key (ls_cli_...) and the Agent Analytics tracking key (ls_track_...) must be created manually via the hosted dashboard (Settings > API Keys), which requires human login/signup. The agent tested with a self-invented bogus key against the real hosted endpoint and correctly got 'Invalid API key', then stopped and asked the user to supply a real key. No authenticated operation against the product was ever completed. Event 144: ```text The official quickstart requires an API key from **Settings > API Keys** in the dashboard — shown once, not creatable via API. I have no account, so `examples/02` and the authenticated half of `examples/05` stop at an explicit message instead of guessing. ``` Event 88: ```text /getting-started -> { detail: 'Invalid API key' } /api-reference -> { detail: 'Invalid API key' } ``` Event 108: ```text You need: A Lightsage organization with Visibility or API Performance data. Access to Settings > API Keys in the Lightsage dashboard. ``` #### Hallucinated URLs None identified in this transcript. #### Blockers - **No self-serve path to a real API key**: Both Lightsage credential types (ls_cli_... for the Public API/CLI/MCP, and ls_track_... for Agent Analytics) are only issued through a human dashboard login at Settings > API Keys, with keys shown once and not creatable via any API call. This is a normal authentication requirement (not a product defect), and it stopped the authenticated examples from running for real. Event 108: ```text Open Settings > API Keys . Click Create API key . Copy the key when it appears. Store it in your shell or secret manager. ``` Event 88: ```text LIGHTSAGE_API_KEY is not set. 1. Create an account at https://lightsage.com (Free plan, no card). 2. Open Agent Analytics -> API key. ``` #### Suggested Changes - **Update docs/account/credits.md and docs/account/plans.md to match the live pricing page**: The cached docs pages at lightsage.com/docs/account/credits.md and /docs/account/plans.md describe a 'Pro' plan at $100/month with 1,500 credits and 250 free starter credits, but the live lightsage.com/pricing page shows Free/$0, Scale/$1,000 (19,000 credits), and Enterprise/custom with no Pro tier. Reconcile the two so an agent reading docs does not quote stale pricing; verify by re-fetching both docs/account/credits.md and /pricing and diffing the plan names and figures. Event 134: ```text > [card] Pro plan > > **1,500 credits per month**, included with the \$100/month subscription. ``` Event 98: ```text Free Scale Recommended $1,000 /month 19,000 monthly credits Start 7-day trial Enterprise Custom annual Contact sales ``` - **Return the documented {ok:false, error} shape on invalid-key tracking calls**: The agent-tracker README's TrackResult type documents an {ok, category, agent, skipped, error} response shape, but a real call to the hosted track endpoint with an invalid key returned a raw {detail: 'Invalid API key'} body instead. Align the live /api/v1/agent-tracking/track error response with the documented TrackResult interface in the SDK README, and confirm by re-running trackVisit() with a bad key and checking the resolved object matches the documented fields. Event 88: ```text /getting-started -> { detail: 'Invalid API key' } /api-reference -> { detail: 'Invalid API key' } ``` Event 30: ```text interface TrackResult { ok: boolean; category?: "coding-agent" | "browsing-agent" | "bot" | "human"; agent?: string; skipped?: string; // "not-page-view" if filtered error?: string; } ``` ### Task given to each agent Help me build a simple example using Lightsage. 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 with 200 when requested with Accept: text/markdown. ``` - **Pass** — llms.txt provides an actionable documentation index ```text llms.txt lists docs, blog, resources, auth.md, ai-sitemap and Devtool Arena pages with markdown URLs. ``` - **Pass** — llms.txt provides navigation guidance ```text llms.txt gives fetch guidance, discovery endpoints, excluded surfaces, and per-page markdown URLs. ``` - **Pass** — llms.txt mentions offered API, MCP, and skills ```text llms.txt mentions public API, CLI, MCP server, AgentID sign-in, and Agent Skills. ``` - **Pass** — A compact guide representation exists ```text llms.txt lists .md URLs for blog, resources, pricing, research and auth pages. ``` - **Pass** — A focused guide is directly retrievable ```text Fetched .md guides return clean text/markdown, e.g. MCP testing and agent-led growth articles. ``` - **Pass** — Equivalent instructions fit a token budget ```text Fetched .md guides are 3-4k tokens, well under the 8000-token budget. ``` - **Pass** — Product-docs links survive format changes ```text Homepage Markdown is supported and links to docs, pricing and MCP reference survive. ``` - **Pass** — The compact guide is independently actionable ```text llms.txt gives a clear starting point, key pages, and fetch instructions for agents. ``` - **Pass** — Install and next-step links resolve ```text Homepage links to docs, pricing, and signup all returned HTTP 200. ``` ### Onboarding - **Skipped** — Docs lead to a relevant quickstart ```text No docs quickstart page fetched; llms.txt lists blog/resources, not a product quickstart. ``` - **Skipped** — Installation commands are extractable ```text No Lightsage installation or CLI setup page was fetched for its own product. ``` - **Skipped** — Code examples are available without interaction ```text Fetched pages are marketing articles; no Lightsage API/SDK code examples were retrieved. ``` - **Skipped** — Prerequisites and auth boundaries are explicit ```text auth.md is referenced but not fetched; no prerequisites or auth boundary content observed. ``` ### Pricing - **Pass** — Pricing is readable without interaction ```text Pricing page renders as static Markdown with all tiers and tables readable without interaction. ``` - **Pass** — Prices are stated, not gated ```text Prices stated: Free $0, Scale $100–$10,000/month, Enterprise custom annual. ``` - **Pass** — Pricing units and limits are explicit ```text Credits, per-credit rates, and estimated prompt/eval capacities are explicit per tier. ``` - **Pass** — Agents identify pricing and its assumptions ```text 2 of 3 sessions were judged on pricing; 0 fell short. DeepSeek V4.1 Flash: Quoted the Free ($0), Scale ($1,000/mo for 19,000 credits at $0.0526 each) and Enterprise tiers from the pricing page, with trial limits, top-ups and estimated capacity per workload. (seq 60, seq 141) Qwen 3.8 Max: Gave Free $0, Scale $1,000/mo with 19,000 credits, and custom Enterprise pricing. It listed per-run credit costs (1 or 5), a usage formula, trial and top-up limits, and flagged outdated pricing in the docs. (seq 23, seq 134, seq 144) This behavioural item does not affect the fast grade. ``` ### Activation - **Skipped** — An API reference or OpenAPI spec is reachable ```text No Lightsage API reference or OpenAPI spec fetched; llms.txt only mentions a public API and auth.md. ``` - **Skipped** — An MCP server is documented and well-formed ```text llms.txt says agents get an API key for the public API, CLI and MCP server, but no MCP server docs were fetched. ``` - **Skipped** — A CLI install path is documented ```text Skills page references a Lightsage CLI setup doc, but no CLI install page was fetched. ``` - **Skipped** — SDK packages resolve on their registries ```text No Lightsage SDK or CLI package registry lookup was supplied; only third-party packages appear. ``` - **Pass** — Agent skills are published ```text Lightsage docs page lists three Agent Skills and an npx skills add install command. ``` [Full report data](https://www.ax-check.com/lightsage.com/report.json)