agentleFS
ElasticKnown publisherelastic/elastic-docs-skills/skills/review/docs-check-kibana-oas/SKILL.md

docs-check-kibana-oas

Skill72 starsChanged 5 months agoApache-2.0View on GitHub

Review a Kibana PR for OpenAPI Spec (OAS) and API docs compliance against the Elastic API docs checklist, core guidelines, and Kibana quickstart. Checks availability, summaries, descriptions, parameters, examples, tags, links, defaults, enums, deprecation, and generated YAML correctness. Use when reviewing a Kibana PR that touches API routes or OAS output files, or when asked to check OAS, availability, x-state, or API docs compliance on a PR.

The file

Show its rules and text

We split this skill into the rules it gives an agent

24 rules. About 0 are common, 12 are distinctive, and 12 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 12

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If no PR number is given, ask for one.

    line 50
  2. Distinctive:

    If no local clone is found at either location, tell the user and ask them for the path to their Kibana checkout.

    line 80
  3. Distinctive:

    Read the actual source files from this checkout for exact line numbers.

    line 82
  4. Distinctive:

    When the PR adds a new route to an existing resource, read the existing routes from the checked-out branch to compare conventions.

    line 102
  5. Distinctive:

    Every file reference must be a clickable markdown link into the PR's "Files changed" view:

    line 110
  6. Distinctive:

    If you cannot compute the hash, link to the PR files page with the bare path.

    line 124
  7. Distinctive:

    Always include the before → after when the fix is a one-liner

    line 163
  8. Distinctive:

    Every file reference is a markdown link to the PR files view

    line 164
  9. Very specific:

    Every new route MUST have availability in its options block with both stability and since.

    line 174
  10. Very specific:

    Every new route MUST have a summary.

    line 209
  11. Very specific:

    Every new route MUST have a description.

    line 227
  12. Very specific:

    Use sentence case: 'oas-tag:Agent builder' not 'oas-tag:agent builder'

    line 266
  13. Distinctive:

    Use consistent tag names across related routes

    line 267
  14. Very specific:

    Every path parameter MUST have meta: { description: '...' } in its schema definition.

    line 277
  15. Very specific:

    Every new request body, query, or response property MUST have meta: { description: '...' }.

    line 294
  16. Very specific:

    When a property uses schema.oneOf or schema.literal to define a fixed set of values, each value should be described, either in the property …

    line 308
  17. Very specific:

    When the default is set server-side (not in the schema), document it in the description instead: 'Sort order.

    line 333
  18. Very specific:

    When a new property is added to an existing route's schema, and that property ships in a later version than the route itself, the property …

    line 342
  19. Distinctive:

    Use realistic data, not placeholders

    line 436
  20. Distinctive:

    Write clear summaries (under 45 characters) for each example

    line 437
  21. Very specific:

    Consider adding x-codeSamples for cURL and Console examples

    line 439
  22. Very specific:

    Use kebab-case verb-resource names, for example create-foo, get-foo, delete-foo.

    line 463
  23. Distinctive:

    Keep operation IDs stable across versions of the same route — don't rename gratuitously.

    line 464
  24. Very specific:

    Use the YAML x-state values to detect problems:

    line 472

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one or more contributor license agreements. See the NOTICE file distributed with this work for additional information regarding copyright ownership. Elasticsearch B.V. licenses this file to you under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. -->

Check Kibana OAS

You are reviewing a Kibana PR from elastic/kibana against the Elastic API docs checklist, core guidelines, and Kibana quickstart. Output a concrete action list. Every item names the file and line to change.

Input

$ARGUMENTS is a PR number or URL. Extract the PR number. Examples:

  • /docs-check-kibana-oas 289719
  • /docs-check-kibana-oas https://github.com/elastic/kibana/pull/289719
  • /docs-check-kibana-oas https://github.com/elastic/kibana/pull/289719/changes

If no PR number is given, ask for one.

Key principle: YAML is generated, TypeScript is the source of truth

The files under oas_docs/output/ (kibana.yaml, kibana.serverless.yaml) are auto-generated from TypeScript route definitions. CI regenerates them on push. Use the YAML diff to detect problems (missing x-state, wrong stability label, empty values), but every action you output must point at the TypeScript source that generates it. Never suggest manually editing the YAML.

Workflow

1. Fetch the PR

gh pr diff <PR_NUMBER> --repo elastic/kibana
gh pr view <PR_NUMBER> --repo elastic/kibana --json title,body,files,headRefName

Save the full diff. You will scan it multiple times.

2. Check out the PR branch

Find a local Kibana clone. Check, in order:

  1. Current working directory (if it is inside a kibana repo)
  2. ~/elastic/kibana

Then check out the PR branch:

cd <KIBANA_DIR> && gh pr checkout <PR_NUMBER>

If no local clone is found at either location, tell the user and ask them for the path to their Kibana checkout.

Read the actual source files from this checkout for exact line numbers. Never compute line numbers from diff hunk headers.

3. Classify the changes

From the diff, sort changed files into:

  • Route definitions: .ts files under server/routes/ registering router.versioned.* endpoints
  • Schema definitions: .ts files with schema.object / schema.maybe / etc. used in route validation
  • Generated YAML: oas_docs/output/ — scan for diagnostic signals only
  • Example files: YAML under routes/examples/
  • Other: types, tests, mocks — note but do not audit

4. Audit

Apply every rule from the rules section below against the diff. Use the generated YAML as a cross-check signal. If the YAML looks wrong (missing x-state, empty x-state, wrong label), trace it back to the TypeScript source to find the root cause.

For every finding, read the actual file from the checked-out branch to confirm the exact line number. Point at the line where the developer needs to make the change.

5. Cross-check sibling routes

When the PR adds a new route to an existing resource, read the existing routes from the checked-out branch to compare conventions. Look for patterns like narrative doc links, tag usage, or example files that siblings use but the new route does not.

6. Output

Print a flat action list, then a summary. Be terse. No diagnosis essays.

Linking to files

Every file reference must be a clickable markdown link into the PR's "Files changed" view:

https://github.com/elastic/kibana/pull/<PR_NUMBER>/files#diff-<SHA256_OF_PATH>

Compute the anchor:

echo -n "<full-repo-relative-path>" | shasum -a 256 | cut -c1-64

Link like: [agents.ts:689](https://github.com/elastic/kibana/pull/290353/files#diff-<hash>)

If you cannot compute the hash, link to the PR files page with the bare path.

Format

## OAS review — PR #<NUMBER>

**<PR title>**

### Actions

1. ❌ **Add `stability` to availability** — [agents.ts:689](<link>)
   `availability: { since: '9.5.0' }` → `availability: { stability: 'experimental', since: '9.5.0' }`

2. ❌ **Add description for `user_id` path param** — [agents.ts:140](<link>)
   Path parameters must have `meta: { description: '...' }`.

3. ❌ **Add narrative doc link** — [attachments.ts:105](<link>)
   Sibling conversation routes link to the agent chat docs. Consider adding a similar link.

4. ❌ **Summary exceeds 45 characters** — [agents.ts:684](<link>)
   `'Get all agent builder conversation attachments for a user'` → shorten to ~40 chars.

### Passes

- ✅ `summary` and `description` present — [agents.ts:684](<link>)
- ✅ Path params have `meta.description` — [agents.ts:700](<link>)
- ✅ Tags include `oas-tag` — [agents.ts:680](<link>)
- ✅ Example file referenced — [agents_get.yaml](<link>)

### Verdict

**N actions** (X required, Y suggested)

Rules:

  • ❌ for actions the author needs to address
  • ✅ for passes — list briefly, one line each
  • Tag pre-existing issues as "(pre-existing)" so the PR author knows what they introduced vs. inherited
  • Always include the before → after when the fix is a one-liner
  • Every file reference is a markdown link to the PR files view

Rules

These rules come from the Elastic API docs checklist, the core guidelines, the organize and annotate guide, and the Kibana quickstart. They also incorporate conventions observed across existing Kibana routes.

1. Route-level availability (lifecycle status)

Every new route MUST have availability in its options block with both stability and since. This powers the version badges and tech preview labels in the published API docs.

Valid stability values and their rendered labels:

ValueRendered labelMeaning
'experimental'ExperimentalCan change or be removed in future versions
'tech_preview'Technical previewPre-release, may change
'stable'Generally availableStable for production use (default if omitted, but be explicit)

since is a version string like '9.2.0'. It marks the version when the API first shipped. It appears in Elastic Stack docs and is omitted from serverless docs.

Correct pattern:

router.versioned.get({
  path: '...',
  options: {
    tags: ['...', 'oas-tag:Agent builder'],
    availability: {
      stability: 'experimental',
      since: '9.6.0',
    },
  },
})

What to flag:

  • ❌ availability block missing entirely
  • ❌ stability missing from availability
  • ❌ since missing from availability
  • ❌ stability has an invalid value

2. Route summary

Every new route MUST have a summary. Summaries appear in IDEs, search results, and documentation overviews.

Summary guidelines:

  • 5–45 characters — keep it short because space is limited in many contexts
  • Start with a verb — "Get", "Create", "Update", "Delete"
  • Use basic verbs — "Get" not "Retrieve", "Update" not "Modify"
  • Include articles — "Delete a space", "Delete spaces"
  • Sentence case — capitalize only the first word and proper nouns
  • No trailing period

What to flag:

  • ❌ summary missing
  • ❌ Summary exceeds 45 characters
  • ❌ Summary does not start with a verb
  • ❌ Summary uses title case or ends with a period

3. Route description

Every new route MUST have a description. Descriptions support markdown formatting and should cover:

  • Purpose and impact: what does this operation do and why would a user need it?
  • Prerequisites or context: what should users know before calling this endpoint?
  • Constraints: valid values, formats, size limits, rate limits
  • Relationships: how parameters interact, how multiple values are handled

A good description adds detail beyond the summary. A bad description just restates it.

Good: 'Create a new conversation with an agent. The conversation persists across sessions and can be shared with other users via access control. To learn more, refer to the [agent chat documentation](https://www.elastic.co/docs/...).'

Bad: 'Creates a conversation.' (restates the summary, adds nothing)

What to flag:

  • ❌ description missing
  • ❌ Description is just a copy or restatement of the summary
  • ❌ Description omits constraints or prerequisites that a user would need

4. Narrative documentation link

Route descriptions should include a trailing sentence linking to the relevant narrative docs page. This is a convention observed across existing Kibana routes. It is most important when sibling routes on the same resource already include such a link.

Pattern:

description:
  'List all conversations. To learn more, refer to the [agent chat documentation](https://www.elastic.co/docs/...).',

How to check: compare against sibling routes. If other routes on the same resource include a "To learn more..." or "refer to the docs" link and the new route does not, flag it.

What to flag:

  • ❌ Missing narrative doc link when siblings include one (warning, not error — the link might not exist yet for a brand-new feature)

5. Tags

Routes MUST have at least one oas-tag: entry in their tags array. Tags group APIs by feature in the published docs.

Tag guidelines:

  • Use sentence case: 'oas-tag:Agent builder' not 'oas-tag:agent builder'
  • Use consistent tag names across related routes
  • One oas-tag per route for clean navigation

What to flag:

  • ❌ No oas-tag:* in the tags array
  • ❌ Tag name uses inconsistent casing vs. sibling routes
  • ❌ Multiple oas-tag: entries on a single route

6. Path parameter descriptions

Every path parameter MUST have meta: { description: '...' } in its schema definition.

Correct pattern:

params: schema.object({
  conversation_id: schema.string({
    meta: { description: 'The unique identifier of the conversation.' },
  }),
}),

What to flag:

  • ❌ A path parameter has no meta or no description inside meta

7. Property descriptions

Every new request body, query, or response property MUST have meta: { description: '...' }.

A good property description explains what the value controls, its format, and any constraints:

  • ❌ 'The page size.' — vague, no constraints, no default
  • ✅ 'The maximum number of results to return. Must be between 1 and 1000. Defaults to 20.'

What to flag:

  • ❌ New property has no meta at all
  • ❌ meta exists but has no description
  • ❌ Description is a single generic phrase that restates the property name (for example, 'The name.' on a property called name)

8. Enum value descriptions

When a property uses schema.oneOf or schema.literal to define a fixed set of values, each value should be described, either in the property description or in the individual schema options.

Skip documenting values that are self-explanatory:

  • true / false, asc / desc, enabled / disabled — obvious from the name
  • read / write / admin — not obvious; what does admin grant beyond write?
  • low / medium / high — not obvious; what concretely changes at each level?

What to flag:

  • ❌ Enum values with no descriptions when the meaning or behavioral difference is unclear from the name

9. Default values

Optional parameters and properties should document their defaults. In Kibana route schemas, this means using schema.maybe(schema.string({ defaultValue: '...' })) or documenting the server-side default in the description.

Correct pattern:

page_size: schema.maybe(
  schema.number({
    defaultValue: 20,
    meta: { description: 'Number of results per page. Defaults to 20.' },
  })
),

When the default is set server-side (not in the schema), document it in the description instead: 'Sort order. The server defaults to descending if not specified.'

How to check: compare against sibling routes on the same resource. If a sibling documents its page_size default but the new route does not, flag it.

What to flag:

  • ❌ Optional parameter with an undocumented default when the behavior changes based on the value

10. Property-level availability

When a new property is added to an existing route's schema, and that property ships in a later version than the route itself, the property MUST have its own meta.availability with stability and since.

How to detect: look for added schema.maybe(...) or schema.object(...) fields inside an existing route's validation block. If the route has since: '9.2.0' and the new property ships in '9.6.0', the property needs its own availability.

Correct pattern:

new_field: schema.maybe(
  schema.string({
    meta: {
      availability: { stability: 'experimental', since: '9.6.0' },
      description: 'Description of the new field.',
    },
  })
),

What to flag:

  • ❌ New property on an older route has no meta.availability
  • ❌ Property has availability but is missing stability or since

11. Deprecation

Deprecated routes or properties should be marked. In Kibana, the route's options supports a deprecated flag, and property descriptions should explain the deprecation.

What to flag:

  • ❌ A property or route is described as deprecated in comments or description text but has no deprecated: true marker
  • ❌ Deprecated property has no description explaining what to use instead

12. Response examples

New routes SHOULD have an example file referenced via oasOperationObject. Examples significantly improve API documentation usability.

Inline TypeScript examples (type-checked at dev time):

.addVersion({
  version: '2023-10-31',
  options: {
    oasOperationObject: () => ({
      requestBody: {
        content: {
          'application/json': {
            examples: {
              example1: {
                summary: 'An example request',
                value: { name: 'Example' } as MyType,
              },
            },
          },
        },
      },
    }),
  },
})

YAML file examples:

import path from 'node:path';
.addVersion({
  version: '2023-10-31',
  options: {
    oasOperationObject: () => path.join(__dirname, 'examples/my_route.yaml'),
  },
})

Example YAML structure (in server/routes/examples/):

requestBody:
  content:
    application/json:
      examples:
        example1:
          summary: Example request
          description: An example of creating a resource
          value:
            name: 'Example'
responses:
  200:
    content:
      application/json:
        examples:
          success:
            summary: Successful response
            value:
              id: '12345'
              name: 'Example'

Example guidelines:

  • Use realistic data, not placeholders
  • Write clear summaries (under 45 characters) for each example
  • Include at least one success response example (HTTP 200)
  • Consider adding x-codeSamples for cURL and Console examples

What to flag:

  • ❌ No oasOperationObject reference (warning — strongly recommended)
  • ❌ Example file contains placeholder or empty values
  • ❌ No response example for the success case

13. Operation ID (advisory only — do not include in the Actions list)

Public routes can optionally set an explicit operationId. SDK and CLI generators use this ID as the method/command name; if omitted, the generator derives one from the method and path (for example, post-foo), which is often unclear.

Correct pattern:

router.versioned.post({
  path: '/api/foo',
  access: 'public',
  summary: 'Create a foo resource',
  operationId: 'create-foo',
  options: { ... },
});

Guidelines:

  • Use kebab-case verb-resource names, for example create-foo, get-foo, delete-foo.
  • Keep operation IDs stable across versions of the same route — don't rename gratuitously.

This is a suggestion, not a compliance rule — do not add it to the ❌ Actions list or count it against the verdict. Only raise it as a one-line side note ("Consider adding an explicit operationId") when the user specifically asks about SDK/CLI generation, operation IDs, or when a route in the diff already sets one inconsistently with the kebab-case verb-resource convention.

14. Cross-check: generated YAML as a diagnostic signal

The YAML files under oas_docs/output/ are auto-generated. Never suggest editing the YAML directly. Every fix goes in the TypeScript source.

Use the YAML x-state values to detect problems:

YAML x-stateTypeScript meaning
Experimental; added in 9.6.0availability: { stability: 'experimental', since: '9.6.0' }
Technical Preview; added in 9.2.0availability: { stability: 'tech_preview', since: '9.2.0' }
added in 9.0.0 (no label)availability: { stability: 'stable', since: '9.0.0' }
Experimental (no version)Normal for kibana.serverless.yaml — serverless omits added in
'' (empty string)Missing stability in the TypeScript availability
Missing entirelyMissing availability block in the TypeScript

What to flag (always pointing at the TypeScript file):

  • ❌ Empty or missing x-state → trace to the TS route and flag missing availability or stability
  • ❌ YAML label does not match TS stability → flag the TS mismatch
  • The serverless YAML omitting added in is expected, not a finding

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.