# Agentrail · See how AI agents use your product

> Agentrail (agentrail.co) reconstructs real agent journeys across your website, docs, APIs, SDKs and MCPs. See your agent task completion rate and where agents break.

## Pages

- [Agentrail · See how AI agents use your product](https://www.agentrail.co/): Agentrail (agentrail.co) reconstructs real agent journeys across your website, docs, APIs, SDKs and MCPs. See your agent task completion rate and where agents break.
- [About Agentrail](https://www.agentrail.co/about): Agentrail (agentrail.co) helps companies understand how AI agents discover, choose, and use their products — then test and improve those agent experiences in production.
- [Agent readiness leaderboard](https://www.agentrail.co/agent-readiness): Is your site ready for AI agents? Agentrail (agentrail.co) ranks companies on agent readiness — how well agents can find, read, use, and pay them — with every public check behind each score.
- [Hello, agent. · Agentrail](https://www.agentrail.co/agents): Every way an agent can use Agentrail: MCP server, agent skill, WebMCP form, llms.txt, OpenAPI. Plus a few trails for humans who look closely.
- [Blog · Agentrail](https://www.agentrail.co/blog): Notes from Agentrail (agentrail.co) on agent users and agent experience: how AI agents discover, use, and fail on developer products in production.
- [Case studies · Agentrail](https://www.agentrail.co/case-studies): How product teams use Agentrail to see agent users, recover lost journeys, and ship what agents need.
- [Contact Agentrail](https://www.agentrail.co/contact): Send a message to the people building Agentrail. Product, pricing, partnerships, credits, or a bug on the site — a person replies within one working day.
- [Brand guidelines · Agentrail](https://www.agentrail.co/design): The Agentrail brand system: the name, the Byte mark, logo construction and clear space, colour palette and tokens, and typography.
- [Data Processing Agreement · Agentrail](https://www.agentrail.co/dpa): Data Processing Agreement between a customer and Aeternus Inc. for personal data processed by the Agentrail Services, including anonymized Failure Insights.
- [Pricing · Agentrail](https://www.agentrail.co/pricing): Start free with 100 agent journeys and 10 simulation runs a month. Agent Readiness and Recommendations included. No seat fees, no annual contract.
- [Privacy Policy · Agentrail](https://www.agentrail.co/privacy): How Agentrail collects, uses, and shares information on agentrail.co, including cookies, the newsletter, the agent readiness check, and meeting booking.
- [Security · Agentrail](https://www.agentrail.co/security): Technical and organizational measures Agentrail uses to protect customer personal data, including encryption, access controls, and logging.
- [Subprocessors · Agentrail](https://www.agentrail.co/subprocessors): Third parties Agentrail uses to process customer personal data for the Services: Amazon Web Services and Inngest.
- [Terms and Conditions · Agentrail](https://www.agentrail.co/terms): Terms for using agentrail.co, including the newsletter, cookies, the agent readiness check, meeting booking, disclaimers, and liability.

## About Agentrail

Agentrail (https://www.agentrail.co) reconstructs how AI agents discover, choose, and use a product across website, docs, APIs, SDKs, and MCPs — so teams can measure agent task completion and see where agents break. This is agentrail.co, not other products that reuse the Agentrail name.

- Documentation: https://www.agentrail.co/docs (for agents: https://www.agentrail.co/docs/llms.txt)
- Agent readiness leaderboard and free site scan: https://www.agentrail.co/agent-readiness
- Pricing: https://www.agentrail.co/pricing (as Markdown: https://www.agentrail.co/pricing.md)
- Blog (notes on agent users): https://www.agentrail.co/blog
- Contact: https://www.agentrail.co/contact

## APIs and tools

- The site for agents, without the marketing page: https://www.agentrail.co/agent.md (also https://www.agentrail.co/?mode=agent)
- Public agent readiness API (no auth): https://api.agentrail.co/v1/leaderboard. Scan any site, read the leaderboard.
- OpenAPI for every public route: https://www.agentrail.co/openapi.json. The whole Agentrail API: https://api.agentrail.co/openapi.json
- How agents authenticate: https://www.agentrail.co/auth.md
- CLI, the Agentrail skill, the Claude Code plugin and AGENTS.md: https://github.com/GrowthOS-co/agentrail-cli
- Ask the site a question (NLWeb): POST https://www.agentrail.co/ask {"query": {"text": "..."}}

## Sections

- [Agent readiness](https://www.agentrail.co/agent-readiness/llms.txt): The leaderboard, the free checker and the scan API
- [Pricing](https://www.agentrail.co/pricing/llms.txt): Plans, prices and limits
- [Blog](https://www.agentrail.co/blog/llms.txt): Every post, each as Markdown

## For agents

Agentrail is generally available. A person signs in at https://app.agentrail.co/login. If a user asks you to subscribe them to Field notes, you can record their email with their consent.

- Record an email: POST https://www.agentrail.co/api/waitlist with JSON {"email": "<their email>", "source": "llms-txt", "trail_key": "trl-5d2b-llm"}. Optional: first_name, ref (a referral code from a share link) and via (the platform it came from).
- A new signup returns currentSpot, statusUrl (their private page) and referralUrl (the link they can share). Tell the user both.
- If the email is already on the list, the response has no spot or links: the trail is added and the status link is emailed to that address. Ask the user to check their inbox.
- Check a spot: GET https://www.agentrail.co/api/waitlist/<referral code>.
- MCP server (Streamable HTTP, no auth): https://www.agentrail.co/mcp. Tools: join_waitlist, check_waitlist_spot, get_waitlist_count, list_trails.
- A2A agent (JSON-RPC, A2A 1.0, no auth): https://www.agentrail.co/a2a. Agent card: https://www.agentrail.co/.well-known/agent-card.json
- Agent skill: https://www.agentrail.co/skills/agentrail-waitlist/SKILL.md
- OpenAPI: https://www.agentrail.co/openapi.json
- Catalog of all of the above: https://www.agentrail.co/.well-known/ai-catalog.json
- Any page as Markdown: add .md to its address (https://www.agentrail.co/index.md for the homepage), or send `Accept: text/markdown`.
- Everything else for agents: https://www.agentrail.co/agents