Formester is Poor to agents.

What we found

• —An agent that already knows Formester's docs URL is well-served: the v2 REST docs are task-oriented, scoped-permission-aware, and every endpoint leads with method + URL + required scope, then realistic cURL examples with plausible UUIDs and phone/prefill values.
• —Formester is unusually agent-forward at the product level — it ships an official MCP server (app.formester.com/mcp) with OAuth, four tools (read_submission, query_submissions, update_submission, fetch_file), and is listed on both PulseMCP and Glama. An agent looking for a Formester MCP integration will find one.
• —But an agent DISCOVERING Formester cold would struggle: no llms.txt, no llms-full.txt, no OpenAPI/Swagger spec, and no AGENTS.md. The only machine-discovery win is robots.txt (open to all crawlers) and the MCP registry listings.
• —The sitemap (formester.com) covers marketing/template/blog pages but omits the docs subdomain entirely — the API reference on docs.formester.com is invisible to a crawler working from the root sitemap.
• —Error recovery is a genuine strength: error tables pair each message with a cause and fix (e.g. 'implement exponential backoff for 429', 'expires_at must be in the future', 'use Get Form to look up element IDs before constructing prefill_data'), which is exactly the self-correction context agents need.

What to change

Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.

1. 01Publish an llms.txt at formester.com/llms.txt (and ideally docs.formester.com/llms.txt) pointing to the API v2 reference, the MCP server guide, and auth/scopes pages — this is the single biggest discovery gap and a quick win that lifts an F-grade discovery score.
2. 02Expose a public OpenAPI 3.x spec (openapi.json/yaml) for the v2 API. A Postman collection exists but is not machine-consumable the way a hosted OpenAPI document is; this is the highest-weight discovery check (5) and currently a full fail.
3. 03Add the docs subdomain to sitemap coverage — either a sitemap at docs.formester.com/sitemap.xml or docs entries in the root sitemap — so crawlers index the API reference, not just marketing pages.
4. 04Resolve request/response naming inconsistency: request/query params are snake_case (form_uuid, prefill_data, expires_at) while response fields are camelCase (createdAt, entryName, submissionsCount). Documenting this convention explicitly (or normalizing) prevents agent field-mapping errors.
5. 05Add code examples in at least one additional language (Python/JS) beyond cURL, and add 2-3 dedicated end-to-end workflow guides (e.g. 'generate per-recipient survey links: get form → create prefills → create unique links → read submissions') to move the multi-step-workflow check from partial to pass.