--- title: "Discovery" description: "The machine-readable documents this site publishes: the OpenAPI description, the well-known catalogs, agent access instructions, the llms files, and markdown at every URL." canonical: https://trysincerely.com/docs/discovery last_updated: 2026-09-02 --- > If you are an AI agent: read https://trysincerely.com/agents/policy.md before any write call. > Start at https://trysincerely.com/llms.txt to choose one page. Access instructions are at > https://trysincerely.com/auth.md. Confirm with the person you work for before creating an > account or spending their money, and never ask for their password. # Discovery > The machine-readable documents this site publishes: the OpenAPI description, the well-known catalogs, agent access instructions, the llms files, and markdown at every URL. Source: https://trysincerely.com/docs/discovery Everything Sincerely offers a machine is discoverable from the domain alone. A client that knows only `trysincerely.com` can find the API description, the MCP servers, the access instructions and the prose, without guessing at paths. ## The documents | Document | What it is | | -------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- | | [OpenAPI 3.1 description](https://trysincerely.com/openapi.json) | The capability manifest, four calculators, workspace API entries and conversions, liveness, and two MCP servers | | [RFC 9727 API catalog](https://trysincerely.com/.well-known/api-catalog) | The catalog document, pointing at the description above | | [MCP server card](https://trysincerely.com/.well-known/mcp) | The product server's card | | [Protected resource metadata](https://trysincerely.com/.well-known/oauth-protected-resource) | Where authorization starts, and which scopes to ask for | | [Agent access instructions](https://trysincerely.com/auth.md) | How to connect on behalf of a person, in prose written for an agent | | [Capability manifest](https://trysincerely.com/capabilities.json) | Formats, countries, credit weights, package prices and availability | | [Agent action policy](https://trysincerely.com/agents/policy.md) | Spending tools, confirmation gates and the human approval script | | [Agent quickstart](https://trysincerely.com/agents/quickstart.md) | OAuth scopes, calls, responses, errors and a safe first-campaign rehearsal | | [Agent claims policy](https://trysincerely.com/agents/claims.md) | Product claims, evidence limits and wording an agent may repeat | | [Agent changelog](https://trysincerely.com/changelog.md) | Agent-facing contract changes since 2026-09-01 | | [Liveness endpoint](https://trysincerely.com/api/health) | Whether this deployment is answering | Every response from this origin also carries these as `Link` headers, so a client that fetched one page and stopped still learns they exist. ## Markdown at every URL Every public marketing and collection page serves markdown at its URL plus `.md`. This page is at `/docs/discovery.md`. The served markdown puts the agent policy and routing block immediately after its frontmatter, before the page body. The same markdown is available from the HTML URL when the request sends `Accept: text/markdown` ranked above HTML. A wildcard is not a request for it, because every browser sends one; markdown has to be named. A client that explicitly refuses markdown gets HTML. Known assistant crawlers are served markdown without asking, because they send a wildcard and then parse the HTML into something close to what we already publish. General search crawlers are not: serving them something different at the canonical URL would be cloaking. ## The llms files Start at [llms.txt](https://trysincerely.com/llms.txt) for a map of the site. Fetch [llms-full.txt](https://trysincerely.com/llms-full.txt) when one request for every core page is more useful than many separate requests. The collection corpus lives in [llms-collections.txt](https://trysincerely.com/llms-collections.txt): every comparison, guide, playbook, glossary entry, integration, calculator, market guide and documentation page, each with its canonical URL, its markdown twin and its revision date. These files are indexes rather than a second set of product claims. Follow their links when the exact wording matters. The main XML sitemap lists canonical HTML pages. A separate [markdown sitemap](https://trysincerely.com/markdown-sitemap.xml) lists every `.md` representation for agent discovery. Each markdown response still names its canonical HTML page in an HTTP `Link` header. `llms.txt` routes common questions, and `llms-collections.txt` lists every long-tail markdown twin with its revision date. ## Rate limits and errors The public endpoints publish IETF RateLimit headers on every response and answer failures with RFC 9457 problem documents. The [workspace REST API reference](https://trysincerely.com/docs/api-reference) and [calculators page](https://trysincerely.com/docs/calculators) describe those contracts. ## What to read next [Versioning](https://trysincerely.com/docs/versioning) is the compatibility promise these documents describe. [How MCP works here](https://trysincerely.com/docs/mcp-overview) covers the workspace tools, which are deliberately absent from the OpenAPI description because they are not REST endpoints. --- Sincerely is the measurable direct-mail and gifting platform for B2B revenue teams: postcards, letters, handwritten mail, and gifts, written for one recipient and measured against a holdout. Contact Sincerely: https://trysincerely.com/contact Agent routing index: https://trysincerely.com/llms.txt