← Back to leaderboard
Vercel API
58
D
Vercel scores 58, a D. Semantics (91), intent (86), error handling (80), and parameters (80) are all strong. Examples are zero across the board, which sinks the overall score because examples carry the largest weight in our framework. The spec also still carries version `0.0.1`, suggesting it is machine-generated and not actively curated. Without examples, agents have to guess at request shape every time even though everything else about the spec is in good shape.
Category breakdown
Examples0
Semantics91
Intent86
Error Handling80
Parameters80
Pagination45
Key findings
- All 304 operations and 69 schemas lack examples
- 12 of 304 operations have missing or short descriptions
- 19 of 69 schemas lack descriptions
- 20 of 304 operations have missing or poor operationIds
- Pagination missing on multiple list operations
What Vercel should fix
- Add at least one example per operation. Vercel's deployment objects, env vars, and project configs have specific shapes that examples make instantly clear.
- Bump the spec version. `0.0.1` signals to AI agents (and humans) that the spec isn't treated as production-quality documentation.
- Add pagination parameters to the 16 list operations missing them — particularly deployments and logs, where agents frequently need to iterate.
- Fix the 20 weak operationIds.
How does your API score?
Run AgenticScore on your own OpenAPI spec — find issues before LLMs do.
npx agenticscore score ./openapi.yaml
Install the CLI →