AgentLens documentation

The scan and the audit — read this first

AgentLens offers two different things, and refuses to blur them. The pre-check is a static scan: ~42 checks against the surface your site presents to an agent, scored in seconds. It is evidence about your surface — not a measurement of behaviour. The audit is that measurement: we run real coding agents (Claude Code, cursor-agent) against your product repeatedly and report what actually happened, with 95% Wilson confidence intervals and the evidence behind every run.

The four layers

Checks are grouped into layers, weighted 20/30/40/10. Checks that do not apply to a site (no MCP server to probe, no commerce signals) are excluded from the denominator entirely — a site is never punished for what it is not.

Discovery — 20 points

Can an agent find you and your documentation at all? Homepage title and meta description quality, robots.txt existence and whether it blocks agent crawlers (ClaudeBot, GPTBot, PerplexityBot and friends), sitemap presence, docs and API links findable from the homepage.

Accessibility — 30 points

Once an agent asks for a page, does it get usable content? Whether the homepage answers agent user-agents with the same status it gives browsers, how much readable text is served without JavaScript, whether llms.txt exists, parses as markdown, and its links resolve, JSON-LD structured data, and whether your docs render without a JS runtime.

Usability — 40 points

Can an agent actually operate the product? A published and valid OpenAPI document with declared auth schemes and typed error responses, machine-readable errors on nonexistent paths, agents.md, and — where advertised — a working MCP server: the initialize handshake, listed tools, and complete input schemas (every tool described, every parameter typed and described).

Payments — 10 points

Can an agent transact? Published x402 / AP2 / ACP markers and 402 challenge behaviour. Nearly every site today scores n/a here, which is the honest answer — the layer can only raise a score, never sink one by default.

Scoring rules

• A passing check earns its full weight; a warning earns half; a failure earns nothing.
• N/A checks leave the denominator — they neither punish nor credit.
• Layer scores renormalise over the layers that had anything applicable.
• Grades: A+ ≥ 95 · A ≥ 86 · B ≥ 70 · C ≥ 48 · D ≥ 28 · F below. An unreachable site scores 0/F and says why.
• Every failing or warning check ships a concrete fix, written to stand alone.

What the audit adds

The pre-check grades your surface. The audit measures outcomes: we run real agent sessions against your MCP server, SDK, or CLI in isolated sandboxes — every JSON-RPC message recorded, every run independently asserted — and report per-scenario, per-client success rates with downloadable evidence. It routinely finds what no static scan can: tasks that succeed in one agent and fail in another, servers whose tools list cleanly but reject every call, error responses that send agents into retry loops. Test this with real agents →