OKF Builder Reference / Open Knowledge Format v0.1, explained for builders OKF v0.1
OKF OpenKnowledgeFormat

OKF prompt generator

Open Knowledge Format prompt library

OKF prompt generator

Copy an Open Knowledge Format prompt for your app, paste it into an AI coding agent, and have it create a structured `okf/` folder with concept files, metadata, links, and maintainer questions.

Copy-paste prompts AI agent context OKF bundle generator Validator-ready

7

prompt variants for common systems

okf/

folder structure every prompt creates

v0.1

validator guidance used by this page

Prompt tabs

Choose the app shape, then copy the prompt.

Each prompt tells the agent what to inspect, which OKF files to create, how to write frontmatter, where to link related concepts, and how to handle unknown facts.

Search intent

People looking for an `OKF prompt generator`, an `Open Knowledge Format prompt`, or an `AI agent knowledge bundle prompt` usually want one practical thing: a prompt that turns their real project into a valid, reviewable OKF folder.

01 / SaaS app

SaaS app OKF prompt

Generate an OKF bundle that explains product metrics, subscription data, dashboards, and recurring operating reviews.

Folder structure

okf/
|-- index.md
|-- log.md
|-- metrics/monthly-recurring-revenue.md
|-- metrics/churn-rate.md
|-- tables/subscriptions.md
|-- dashboards/revenue-dashboard.md
`-- playbooks/revenue-review.md

Copy-paste agent prompt

text

Create an OKF bundle for my SaaS app.

First inspect the repository, product docs, database schema, analytics notes, and dashboards that are available to you. Then create an `okf/` folder that helps humans and AI agents understand the app's core commercial knowledge.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- metrics/monthly-recurring-revenue.md
|-- metrics/churn-rate.md
|-- tables/subscriptions.md
|-- dashboards/revenue-dashboard.md
`-- playbooks/revenue-review.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use `Metric`, `Table`, `Dashboard`, and `Playbook` types where appropriate. Use the current date for timestamps in ISO 8601 UTC.

Write concise Markdown bodies that explain the concept, where the source lives, how it is calculated or used, known caveats, and related files. Use bundle-relative links such as `/tables/subscriptions.md`. Do not invent business rules, SQL logic, or dashboard behavior. If a fact is unknown, add it to `okf/log.md` under "Questions for maintainers".

When finished, summarize every file created and tell me how to validate the folder with the OKF validator.

02 / Laravel app

Laravel app OKF prompt

Generate an OKF bundle that maps routes, models, policies, jobs, commands, integrations, and operational behavior in a Laravel codebase.

Folder structure

okf/
|-- index.md
|-- log.md
|-- routes/web.md
|-- routes/api-customers.md
|-- models/user.md
|-- policies/customer-policy.md
|-- jobs/sync-stripe-customer.md
`-- commands/rebuild-search-index.md

Copy-paste agent prompt

text

Create an OKF bundle for my Laravel app.

Inspect the Laravel routes, controllers, models, policies, jobs, commands, migrations, config files, tests, and README files before writing. Create an `okf/` folder that explains the application as linked Open Knowledge Format concept files.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- routes/web.md
|-- routes/api-customers.md
|-- models/user.md
|-- policies/customer-policy.md
|-- jobs/sync-stripe-customer.md
`-- commands/rebuild-search-index.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use types such as `Route`, `Model`, `Policy`, `Job`, `Command`, `Integration`, and `Workflow`.

The Markdown body should explain what the concept does, the source files behind it, important dependencies, failure modes, permissions, related database tables, and related OKF files. Link related concepts with bundle-relative links. Do not guess hidden behavior. If the repository does not show something clearly, record the question in `okf/log.md`.

When finished, summarize the created OKF files and note any routes, jobs, or models that still need maintainer confirmation.

03 / WordPress site

WordPress site OKF prompt

Generate an OKF bundle that documents post types, taxonomies, custom fields, templates, plugins, and editorial workflows.

Folder structure

okf/
|-- index.md
|-- log.md
|-- post-types/product.md
|-- taxonomies/product-category.md
|-- acf/product-fields.md
|-- templates/single-product.md
`-- plugins/woocommerce.md

Copy-paste agent prompt

text

Create an OKF bundle for my WordPress site.

Inspect the theme, plugins, custom post types, taxonomies, ACF field groups, templates, shortcodes, editor patterns, and available documentation. Create an `okf/` folder that explains the site structure in Open Knowledge Format.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- post-types/product.md
|-- taxonomies/product-category.md
|-- acf/product-fields.md
|-- templates/single-product.md
`-- plugins/woocommerce.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use types such as `PostType`, `Taxonomy`, `FieldGroup`, `Template`, `Plugin`, and `Workflow`.

Write Markdown that explains what each concept controls, where it is configured, which templates or plugins depend on it, how editors use it, and which related OKF files matter. Use bundle-relative links. Do not invent plugin settings or editorial policies. Put unknowns in `okf/log.md` as maintainer questions.

When finished, summarize the files created and explain how to upload the `okf/` folder to the OKF validator.

04 / API docs

API docs OKF prompt

Generate an OKF bundle around endpoints, schemas, authentication, errors, rate limits, and business constraints.

Folder structure

okf/
|-- index.md
|-- log.md
|-- endpoints/create-customer.md
|-- endpoints/list-invoices.md
|-- schemas/customer.md
|-- errors/rate-limit.md
`-- auth/api-key-authentication.md

Copy-paste agent prompt

text

Create an OKF bundle for my API docs.

Inspect the OpenAPI file, route definitions, controller code, SDK docs, examples, tests, authentication logic, and error handling. Create an `okf/` folder that makes the API understandable as linked Open Knowledge Format concept files.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- endpoints/create-customer.md
|-- endpoints/list-invoices.md
|-- schemas/customer.md
|-- errors/rate-limit.md
`-- auth/api-key-authentication.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use types such as `Endpoint`, `Schema`, `Error`, `Authentication`, `RateLimit`, and `Workflow`.

Each Markdown body should explain the request, response, constraints, source route or spec location, relevant schemas, common errors, and related files. Link endpoints to schemas, auth rules, and errors using bundle-relative links. Do not invent parameters, status codes, or limits. Put missing details in `okf/log.md`.

When finished, provide a short summary of generated files and validation steps.

05 / Data warehouse

Data warehouse OKF prompt

Generate an OKF bundle that describes datasets, tables, metrics, lineage, dashboards, freshness rules, and quality checks.

Folder structure

okf/
|-- index.md
|-- log.md
|-- datasets/sales.md
|-- tables/orders.md
|-- tables/customers.md
|-- metrics/gross-revenue.md
|-- lineage/order-revenue-lineage.md
`-- dashboards/sales-performance.md

Copy-paste agent prompt

text

Create an OKF bundle for my data warehouse.

Inspect the warehouse schema, dbt models, SQL queries, metric definitions, BI dashboards, lineage docs, freshness checks, and data quality tests available to you. Create an `okf/` folder that explains the analytical system as linked Open Knowledge Format concept files.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- datasets/sales.md
|-- tables/orders.md
|-- tables/customers.md
|-- metrics/gross-revenue.md
|-- lineage/order-revenue-lineage.md
`-- dashboards/sales-performance.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use types such as `Dataset`, `Table`, `Metric`, `Lineage`, `Dashboard`, and `QualityCheck`.

The Markdown body should state grain, owners if known, source tables, transformations, metric formulas, freshness expectations, caveats, and related OKF files. Use bundle-relative links. Do not invent SQL, owners, freshness SLAs, or definitions. Put uncertain items in `okf/log.md`.

When finished, summarize the generated bundle and include validation instructions.

06 / Company knowledge

Company knowledge OKF prompt

Generate an OKF bundle for policies, teams, systems, recurring processes, and decision records.

Folder structure

okf/
|-- index.md
|-- log.md
|-- teams/support.md
|-- policies/refunds.md
|-- systems/billing.md
|-- playbooks/incident-response.md
`-- decisions/pricing-packaging.md

Copy-paste agent prompt

text

Create an OKF bundle for my company knowledge base.

Inspect the available docs, README files, runbooks, policies, decision records, team notes, and system documentation. Create an `okf/` folder that turns important organizational knowledge into linked Open Knowledge Format concept files.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- teams/support.md
|-- policies/refunds.md
|-- systems/billing.md
|-- playbooks/incident-response.md
`-- decisions/pricing-packaging.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use types such as `Team`, `Policy`, `System`, `Playbook`, `Decision`, and `Process`.

Write practical Markdown bodies that define the concept, state where the source material came from, identify related systems or owners when visible, and link related OKF files. Do not invent internal policies, owners, or decisions. Add uncertain or missing details to `okf/log.md`.

When finished, summarize the generated files and list the highest-priority maintainer review questions.

07 / AI agent context

AI agent context OKF prompt

Generate an OKF bundle that gives an AI agent durable context before it works with code, tools, APIs, or data.

Folder structure

okf/
|-- index.md
|-- log.md
|-- systems/billing.md
|-- tools/stripe.md
|-- metrics/monthly-recurring-revenue.md
|-- playbooks/support-triage.md
`-- constraints/agent-safety-rules.md

Copy-paste agent prompt

text

Create an OKF bundle for my AI agent context.

Inspect the repository, product docs, tool docs, runbooks, policies, data definitions, API docs, and examples that the agent will need before acting. Create an `okf/` folder that gives the agent durable Open Knowledge Format context.

Use this folder structure as the starting point:

okf/
|-- index.md
|-- log.md
|-- systems/billing.md
|-- tools/stripe.md
|-- metrics/monthly-recurring-revenue.md
|-- playbooks/support-triage.md
`-- constraints/agent-safety-rules.md

For every non-reserved concept file, add YAML frontmatter with `type`, `title`, `description`, `resource`, `tags`, and `timestamp`. Use types such as `System`, `Tool`, `Metric`, `Playbook`, `Constraint`, and `Workflow`.

The Markdown body should explain what the agent must know, what source file or document supports it, which actions are allowed or risky, and which related files should be read next. Use bundle-relative links. Do not invent tool permissions, safety rules, or business logic. Put unknowns in `okf/log.md`.

When finished, summarize the files created and recommend the first validation and maintainer review steps.

How to use it

How do you use an Open Knowledge Format prompt with an AI agent?

Paste the prompt into the agent that can read your repository or documentation, let it create the folder, then review and validate the generated Markdown files.

  1. 1 Choose the prompt that matches the system: SaaS app, Laravel app, WordPress site, API docs, data warehouse, company knowledge, or agent context.
  2. 2 Paste the prompt into an agent with access to the repo, docs, schema, dashboard notes, or knowledge base you want represented.
  3. 3 Ask the agent to create the `okf/` folder, write concept files, link related files, and list unknown facts in `log.md`.
  4. 4 Upload the generated folder to the validator, fix errors first, then improve warnings before sharing the bundle.

What should the prompt create?

A useful OKF prompt should ask for a small but complete bundle: a root index, a log, concept files, metadata, links, citations when needed, and review questions.

Required

Concept frontmatter

Every non-reserved concept file needs a non-empty `type`; strong prompts also request `title`, `description`, `resource`, `tags`, and `timestamp`.

Structure

Folder-level context

Use `index.md` to orient readers and `log.md` to capture changes, assumptions, and questions for maintainers.

Traversal

Bundle-relative links

Links such as `/tables/orders.md` help people, tools, and AI agents move between related metrics, tables, APIs, dashboards, and playbooks.

Trust

No invented facts

The prompt should tell the agent to inspect sources first and record unknown owners, formulas, permissions, or policies as review questions.

Which OKF prompt should you choose?

Next step

Validate the generated folder before using it.

The prompt helps create structure. The validator checks that the generated Markdown can actually be parsed and reviewed as an OKF bundle.

FAQ

What is an OKF prompt generator?

An OKF prompt generator gives you a copy-paste prompt for an AI agent to create an Open Knowledge Format folder for a specific app, repo, API, warehouse, or knowledge base.

What does an Open Knowledge Format prompt create?

An Open Knowledge Format prompt should create an okf/ folder with index.md, log.md, linked Markdown concept files, YAML frontmatter, bundle-relative links, and maintainer questions for unknown facts.

How should I validate an AI-generated OKF bundle?

Upload the generated okf/ folder to the OKF validator, then fix errors first and warnings second. Errors usually block parseability; warnings improve metadata, links, timestamps, and discovery.