Use from AI agents
These docs are written to be read by people and by machines. Everything an agent needs is available as plain text.
Machine-readable docs
| Resource | URL | Use it for |
|---|---|---|
llms.txt |
/llms.txt |
An index of the docs with one line per page |
llms-full.txt |
/llms-full.txt |
Every guide and the API reference in a single file |
| Markdown pages | Add .md to any page URL, for example /quickstart.md |
Reading one page without HTML |
| API reference in Markdown | /reference.md |
Every operation, schema and example |
| OpenAPI 3.1 | /openapi.json · /openapi.yaml |
Generating clients and tool definitions |
All of them are served with Access-Control-Allow-Origin: *.
MCP server
Coming soon. A remote MCP server at mcp.salchimail.com will let an agent verify addresses, run batches and read usage with your API key. This page will link to it when it is available.
Agent skill
Coming soon. A skill that teaches a coding assistant how to integrate SalchiMail. This page will link to it when it is available.
Until both exist, point your agent at /llms.txt or at the OpenAPI contract: it is enough to call the API correctly.
Rules to give your agent
If you write the instructions yourself, include these:
- The API key is a secret. Read it from an environment variable, send it only to
https://api.salchimail.comand never print it. - Build and test with a
sk_test_key: results are simulated and no credits are used (Test mode). - Never treat
unknownasinvalid(Verdicts and reason codes). - For more than a few addresses, create a batch instead of looping over
POST /verify(Clean a list in batches). - Send an
Idempotency-Keyon everyPOST, and reuse it on retries (Idempotency). - Branch on
verdict,reason_codeand the errorcode. Never parsereasonormessage. - A verification never sends email. Do not tell users that a message was sent to the address.