Render is Good to agents.
Discry independently scored how well an AI agent can discover and understand the Render API from what’s public — not whether it’s usable. Below: every signal we checked, what’s costing the score, and what to change.
SCORED UNDER RUBRIC 1.2 · A full re-launch under Discry Score 2.5 — a new behavioral instrument, not comparable to these scores — is in progress.
Discovery
45% of score · 88/100Comprehension
55% of score · 88/100What we found
- Render has one of the most thoughtful llms.txt implementations seen: includes metadata, permissions (train: allow, summarize: allow), naming conventions, citation format, and priority guidance for AI agents — a model for the industry
- The 'Using Render with Coding Agents' dedicated doc page shows explicit investment in agent UX, explaining .md URL appending, Accept header support, and MCP server integration
- AGENTS.md is present in render-oss repos (skills, cli) and the GitHub org has an official MCP server, Skills repo, and CLI — comprehensive agent tooling ecosystem
- robots.txt uses Content-Signal to allow search but disallows ai-train and ai-input — a nuanced position that allows agent access but restricts training use
- llms-full.txt at 825KB provides the entire documentation in a single markdown file with a glossary, link references, and clean structure — exceptional for deep agent comprehension
What to change
Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.
- 01Add .well-known/mcp.json at render.com pointing to the official MCP server at mcp.render.com/mcp — this is the only major discovery signal missing
- 02Document API rate limits explicitly — currently not prominently featured in docs, which means agents cannot plan request pacing
- 03Replace placeholder values in curl examples (e.g., {{render_api_token_goes_here}}) with realistic-looking masked values to help agents understand expected formats
- 04Add dedicated API error recovery documentation showing what to do for common error codes (401, 403, 404, 429, 500) with actionable next steps
Execution coverage · INFORMATIONAL, UNSCORED
Whether an agent can actually complete a call and recover from errors is the deeper Audit layer — documented here, but not part of the Discry Score.
API key authentication via Bearer token. OpenAPI 3.0 spec available at api-docs.render.com. Interactive API reference with multi-language examples. Rate limit details not prominently documented. Pagination supported for list endpoints.