Siteline

About

A Lighthouse score, but for AI agents.

Siteline makes the agentic web scoreable, one public website at a time. Launched on March 19, 2026, it is the earliest working public agent-readiness scanner we could verify. We think it's still the best scanner for evidence-first assessment: it tests whether machine-facing resources actually help agents, records probe-level evidence, and never treats presence alone as proof.

Each scan evaluates whether AI agents can use the submitted public surface on behalf of a human. The result is a bounded website assessment, not validation of agent performance across the web. It scores four pillars — Signal, Navigate, Absorb, and Perform — and assesses dedicated machine-readable resources for agents.

Who it is for

Anyone who wants to know whether their website works for the next wave of web traffic: AI agents browsing on behalf of users, procurement workflows, research assistants, and other machine-mediated interactions.

  • Site owners checking agent readiness before launch
  • Consultants benchmarking client sites against competitors
  • Agencies adding agent usability to their audit toolkit
  • Agent operators evaluating whether a target site is navigable and safe to hand a human through

What it measures

Two independent layers determine the grade. Both must be strong for a site to score well.

  • Layer 1: SNAP Fundamentals — Can agents passively use the site? Signal, Navigate, Absorb, and Perform map to reachability, routing, content quality, and next-step clarity.
  • Layer 2: Agentic Enablement — Does the site provide dedicated machine-readable resources? Eight agentic purposes are assessed for presence and quality, producing a level (0–4) that caps the maximum grade.

What agents can use today

Siteline exposes more than a marketing page. Agents can run scans and interpret results through multiple interfaces.

  • Public REST API for URL-based evaluation
  • Human-facing scanner for operators and clients
  • Internal CLI for maintainers with private repository access
  • Internal stdio MCP server for maintainers; public agents use REST

What it is not

Siteline is not AI SEO, GEO scoring, chatbot optimization, prompt engineering, LLM manipulation, or a protocol-generation tool. It does not guarantee AI traffic or citations. It does not replace a full implementation plan. It makes machine accessibility visible so you know what to fix and where to start.

Roadmap

Siteline is actively maintained for correctness, security, compatibility, and truthful public contracts. New platform investment is parked. Existing services remain available; the items below are future candidates.

  • Expanded report outputs beyond the current public sharing tools
  • Hosted API and remote MCP access
  • Monitoring platform automation and historical comparisons
  • Deeper crawl and protocol-layer advisory work

Constructive boundaries

Siteline depends on the Graceful Boundaries specification for constructive limit and failure communication. Its public API reports Level 4 conformance: limits are discoverable before a request, refusals explain what happened and when to retry, and successful responses include proactive limit headers.

  • Limits discoverable at /api/limits
  • Structured refusal with retry guidance on every 429
  • Proactive RateLimit headers on success responses

Built by PAICE.work PBC

Siteline is built by PAICE.work PBC. We're on a mission to make Aggregate Intelligence measurable, teachable, and governable.

Run a free scan Services FAQ Contact