For API platform, DevEx, and MCP teams

Your API is valid.
Is it agent-ready?

Turn an OpenAPI contract into evidence that agents can discover the right operation, construct it correctly, and interpret the result safely—before it reaches production.

DeterministicVersioned rules, stable results
Bounded retentionRaw specs are not retained
Agent-nativex402 + MCP artifacts
readiness audit · 2026-07-17.2
$ readiness audit inventory.yaml
Agent readiness47/100
needs work
Structure
87
Operation clarity
31
Input schemas
71
Output schemas
58
Error models
0
Agent safety
0
criticalagent.operation.operation-id.missing
POST /inventory/adjustments has no operationId./paths/~1inventory~1adjustments/post

Stable operation identifiers are required for predictable tool names and client generation.

From contract to confidence

A linter checks syntax.
This checks usefulness.

Every result is traceable to a stable rule, an exact JSON Pointer, and a clear impact on agent behavior.

01 · Finding

Find semantic gaps

Catch contracts that validate cleanly but leave agents guessing when, why, and how to call an operation.

criticalPOST /inventory/adjustments
Stable operation ID missingagent.operation.operation-id.missing
02 · Patch

Apply safe repairs

Receive RFC 6902 patches only when a change can be derived mechanically—never invented business semantics.

RFC 6902 · high confidence+ add /paths/~1inventory~1adjustments/post/operationIdpostInventoryAdjustments
03 · Tool

Shape MCP tools

Generate focused candidate tool bundles with collision-safe names, input schemas, and explicit mutation risk.

getOrdereligible

Use this read-only operation when an agent needs the current details for one known customer order.

GET /orders/{orderId} · read only
One deterministic pipeline

From contract to agent evidence.

  1. 01OpenAPI contract
  2. 02Rule findings
  3. 03Safe patches
  4. 04MCP tools
Free score · full sample reports · paid custom audits

See what an agent sees.

Paste, upload, or import a public URL for an OpenAPI 3.0 or 3.1 contract. Raw specifications are processed in memory and not retained; paid derived reports may be cached for up to 24 hours.

Browser-safe URL import
Guided examples

Explore three audit outcomes

Open a precomputed full sample report, or run the free summary before generating a new audit for your own contract.

mutation-needs-work.yamlDemo sample loaded
Audit report preview

Ready when your contract is

Awaiting score
/ 100

Run the free score to populate the report with deterministic evidence from your contract.

01Overall readiness
02Six category scores
03Exact finding locations
04Safe patch groups
05MCP tool candidates
06Integrity fingerprints
Built with your feedback

Help shape what agent-ready means.

Report problems, challenge an audit result, or propose the next readiness rule in our public GitHub community.

Join the discussion →