# JSON Schemas

Describes Keryn 0.5.2 and keryn-rules 0.8.0.

> The JSON Schemas Keryn publishes: configuration, policy, decisions, scope, application, ledger and scan result, with the TypeScript types in @keryn/sdk.

Source: https://keryn.dev/docs/reference/schemas/

Every schema is JSON Schema draft-07 at a stable address. The SDK ships a TypeScript type for each, generated from the same files.

| Schema | Address | What it describes | Type in @keryn/sdk |
|---|---|---|---|
| config | [/schemas/config/v1.json](https://keryn.dev/schemas/config/v1.json) | Keryn project configuration (.keryn.yml) | `KerynConfig` |
| policy | [/schemas/policy/v1.json](https://keryn.dev/schemas/policy/v1.json) | Keryn AI-BOM policy (keryn.dev/policy/v1): Policy-as-code over an AI-BOM inventory: rules (deny / allowOnly / requires over component selectors), time-boxed waivers, `changes` for prompt / model changes found against a base, and `licenses` for the licences of locked dependencies | `KerynPolicy` |
| decisions | [/schemas/decisions/v1.json](https://keryn.dev/schemas/decisions/v1.json) | Keryn decisions (keryn.dev/decisions/v1), .keryn/decisions.yml: The organisation's response to findings, reviewed through pull requests | `KerynDecisions` |
| scope | [/schemas/scope/v1.json](https://keryn.dev/schemas/scope/v1.json) | Keryn scope (keryn.dev/scope/v1), .keryn/scope.yml: What is in scope for a release decision, what is deliberately left out and why, and the coverage the organisation declares | `KerynScope` |
| application | [/schemas/application/v1.json](https://keryn.dev/schemas/application/v1.json) | Keryn application profile (keryn.dev/application/v1), .keryn/application.yml: One application's sources, built artifacts and web targets, so one keryn ci run covers all of them and evidence.json binds the run to this profile's sha256 (bindingBasis profile-declared) | `KerynApplication` |
| ledger | [/schemas/ledger/v1.json](https://keryn.dev/schemas/ledger/v1.json) | Keryn findings ledger (keryn.dev/ledger/v1), ledger.json: One merged, deduplicated, reproducible findings ledger per keryn ci run (or keryn ledger build) | `KerynLedger` |
| result | [/schemas/result/v1.json](https://keryn.dev/schemas/result/v1.json) | Keryn scan result: What `keryn scan --json`, `keryn bom --json` and `keryn decide` write (keryn.dev/result/v1) | `KerynResult` |

The result schema is closed on the fields every consumer reads (`schema`, `versions`, `subject`, `revision`, `coverage`, `components`, `bom`, `findings`, `gate`) and open on engine detail. A test validates real `keryn scan --json` output against it. Point your editor's YAML or JSON plugin at an address for completion and validation.
