Prisma is Agent-Ready to agents.
Discry independently scored how well an AI agent can discover and understand the Prisma 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 · 95/100Comprehension
55% of score · 96/100What we found
- An agent building with Prisma gets clean, structured content at every layer: a well-structured llms.txt, a full llms-full.txt, and a 'Copy Markdown' button on every docs page — so it can ingest documentation without wrestling HTML.
- Prisma ships an official AGENTS.md across its primary repos (prisma/prisma, prisma-engines, language-tools) plus a dedicated prisma/skills repo, with CLAUDE.md and GEMINI.md symlinked to it — coding agents get repo-specific context out of the box.
- Prisma runs its own MCP server (listed on Glama) for natural-language database management, so an agent can provision and query Prisma Postgres directly rather than only reading docs.
- The Prisma Postgres Management API exposes a public, interactive OpenAPI 3.1 spec (api.prisma.io/v1) with a TypeScript SDK, letting an agent discover and call the full endpoint surface programmatically.
- The only real gap: no /.well-known/mcp.json manifest — an agent probing that standard path won't auto-discover the MCP server; it has to find it via the docs or an MCP registry.
What to change
Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.
- 01Add a /.well-known/mcp.json manifest pointing to the Prisma MCP Server so agents can auto-discover it at the standard well-known path — the only failed discovery check.
- 02Strengthen the Error Reference with explicit per-error recovery steps (e.g., 'On P2002, read meta.target for the colliding field and surface a conflict message') rather than error code plus one-line description alone.
- 03Surface the Management API's rate limits and any idempotency semantics explicitly in the API reference so agents can implement backoff and safe retries without discovering limits by failing.
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.
Prisma documents auth thoroughly (OAuth 2.0 + service tokens for the Management API; connection URLs and API keys for the ORM and Prisma Postgres), typed error codes with descriptive `meta` fields for the ORM, and both offset (skip/take) and cursor-based pagination. Rate limits and idempotency semantics are not prominently documented and would require live testing to verify.