Smartlead is Good to agents.

What we found

• —An agent discovering Smartlead lands well: robots.txt on the docs host carries an explicit Content-Signal (ai-train=yes, search=yes, ai-input=yes), and a clean, API-focused llms.txt maps every guide, core concept, and endpoint group — the API is understandable in under ~2,500 tokens before any page is opened.
• —Comprehension is essentially flawless. Endpoint pages lead with the method, URL, and a task-oriented purpose, ship copy-pasteable cURL/Python/JS/PHP examples with realistic values, and consistently use snake_case — an agent can go from zero to a live campaign by following the docs alone.
• —Error recovery is best-in-class for agent use: a dedicated guide pairs each common error (401, 400, 404, 409, 429, 5xx) with an explicit fix, exponential-backoff-with-jitter code, and Retry-After handling — the docs tell an agent what to DO, not just what broke.
• —The single biggest gap is the OpenAPI spec: the Mintlify 'Try it' reference is generated from an underlying schema, but no machine-readable spec is downloadable at any standard path (openapi.json, openapi.yaml all 404) — an agent that wants to auto-generate a client can't fetch one.
• —There is no AGENTS.md in a Smartlead-owned repo, and the vendor-linked .well-known/mcp.json points at a non-canonical Mintlify branch-preview URL (smartlead.main-kill-isr.mintlify.me) rather than a stable production MCP endpoint — a coding agent following it could hit a dead or unstable server.

What to change

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

1. 01Publish a downloadable OpenAPI spec at a discoverable path (e.g. https://api.smartlead.ai/openapi.json) and link it from llms.txt. This is the highest-weighted discovery signal and the only comprehension-adjacent gap — it lets agents auto-generate typed clients instead of scraping HTML.
2. 02Fix the .well-known/mcp.json URL to point at a stable production MCP endpoint (not a Mintlify preview subdomain), so agents that resolve the manifest connect to a durable server.
3. 03Add an AGENTS.md to the primary public Smartlead repo (or publish an official, clearly-branded MCP server repo with one) — the ecosystem already has 6+ community MCP servers, so an authoritative agent-facing entry point would consolidate discovery.
4. 04Document pagination explicitly (cursor vs. offset, page-size limits, response envelope) for the list endpoints (campaigns, leads, inbox) so agents can iterate large result sets deterministically.
5. 05Document idempotency support (or its absence) for write endpoints like campaigns/create and lead import, so agents can safely retry after a 5xx without creating duplicates.