Upstash is Needs Work to agents.
Discry independently scored how well an AI agent can discover and understand the Upstash 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 · 52/100Comprehension
55% of score · 91/100What we found
- Upstash has the strongest agent-resource investment in this cohort: a dedicated /docs/agent-resources section with CLI, MCP server, Skills, and llms.txt documentation — explicitly designed for AI agent consumption
- The llms-full.txt at 2.5MB provides comprehensive markdown content covering every page in the docs — excellent for deep agent comprehension but the primary llms.txt at 115KB is still too large for quick orientation
- Multi-step workflow documentation is exceptional: the Workflow product inherently documents chaining operations with step-by-step guides, retries, and error handling built into the product design
- No OpenAPI specification is publicly available despite having a REST API for Redis, QStash, Vector, and the Developer API — a significant gap for automated agent integration
- The robots.txt allows all bots but does not include explicit AI bot signals like Content-Signal headers
What to change
Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.
- 01Publish an OpenAPI spec for the Developer API (management endpoints for Redis, QStash, Vector databases) — this is the highest-weight discovery check and would significantly boost the score
- 02Add .well-known/mcp.json at the root domain with tool declarations pointing to the official MCP server — connects the strong MCP ecosystem presence to a standard discovery path
- 03Create a concise API-focused llms.txt (<10KB) at the root domain that covers core REST API patterns across Redis, QStash, and Vector — the current 115KB /docs/llms.txt is too large for quick agent orientation
- 04Add AGENTS.md to the upstash/upstash-redis or upstash/mcp-server repos explaining how coding agents should integrate with Upstash services
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.
Bearer token authentication via REST API tokens. Error responses documented with HTTP status codes. Rate limits and flow control extensively documented for QStash. Pagination supported. Dedicated agent-resources section with CLI, MCP server, and Skills documentation.