The best-benchmarked open-source AI memory system. And it's free.
OKF publishing CLI and MCP server
Kiso turns an OKF bundle into a static website with generated HTML, `llms.txt`, and `sitemap.xml`, while keeping the original Markdown files in the output. It also ships an MCP server that exposes concept search and concept content from the same bundle without changing the source.
Builders who keep knowledge in OKF bundles and want to publish it for static websites or agent access.
You can validate one source bundle once and use it both as a public website and as context for MCP-capable agents.
What it does
Validate OKF bundles
`kiso-cli check` scans Markdown for formatting and structural errors and warns about broken links.
Build static websites
`kiso-cli build` generates HTML pages, the original Markdown files, `llms.txt`, and `sitemap.xml`.
Serve bundle knowledge through MCP
`kiso-mcp-server` exposes `search_concepts` and `get_concept_content` over `/mcp` for compatible agent apps.
Use publishing profiles
Profiles in `.kiso/<profile>/configuration.yaml` let you publish the same bundle in different ways without changing source content.
Run in GitHub Actions
`applications/kiso-cli-action` lets you build bundles automatically on pushes.
How to get it
- 1Download the latest release of Kiso for your operating system from the releases page,…
./kiso-cli check --source=examples/kb-google-example
- 2The check command validates the Markdown files in the OKF bundle and reports formatting…
./kiso-cli build --source=examples/kb-google-example --destination=public
README
Kiso
Kiso provides two applications built around Open Knowledge Format (OKF) bundles:
kiso-clivalidates an OKF bundle and publishes it as a static website containing generated HTML pages and the original Markdown files. It can be run from the command line or through GitHub Actions.kiso-mcp-serverstarts a Model Context Protocol (MCP) server that lets an MCP-compatible AI application access the knowledge contained in an OKF bundle.
The OKF bundle remains the single source of truth for both applications.
Kiso CLI
Quick start
If you need an OKF bundle to test Kiso, you can download our sample OKF bundle from the Kiso website. Once downloaded, unzip it.
Download the latest release of Kiso for your operating system from the releases page, then run:
./kiso-cli check --source=examples/kb-google-example
The check command validates the Markdown files in the OKF bundle and reports formatting or structural errors (it also reports broken links as warnings). Once the bundle is valid, build the static website:
./kiso-cli build --source=examples/kb-google-example --destination=public
You will find the generated static website in the public directory. You can open the index.html file in your browser to view the generated site.
If you use the native Kiso binary and want Kiso to generate social network preview images, install one of the supported native SVG to PNG conversion tools: rsvg-convert, inkscape, or resvg.
GitHub Actions
You can use Kiso CLI in GitHub Action to automatically build your OKF bundles into static websites whenever you push changes to your repository. Here's an example workflow configuration:
- name: Build with Kiso
uses: oak-invest/kiso/applications/kiso-cli-action@v0.2.4
with:
command: build
source: examples/kb-google-example
destination: website/examples/kb-google-example-latest
Commands
| Command | Description |
|---|---|
check | Validates the Markdown files in an OKF bundle and reports formatting or structural errors. |
build | Generates a static website, including the original Markdown files, HTML pages, llms.txt, and sitemap.xml. |
Options
| Variable | Command | Default | Description |
|---|---|---|---|
source | check, build | . | Directory containing the Markdown files to read. Defaults to the current directory. |
destination | build | public | Directory where generated files are created. Defaults to the public directory. |
profile | build | Publishing profile loaded from .kiso/<profile>/configuration.yaml. |
Configuration file
The build command can be configured with an optional .kiso/configuration.yaml file located at the root of the source bundle. When the file or one of its sections is absent, Kiso uses safe default values.
site:
baseUrl: https://knowledge.example.com/
name: Example Knowledge Base
language: en
title: My knowledge base
description: Documentation for humans and AI agents
theme:
name: corporate
content:
ignorePatterns:
- README.md
- drafts/**
- private/**
| Property | Default | Description |
|---|---|---|
site.baseUrl | Public base URL prepended to generated site links and sitemap entries. | |
site.name | Name of the site, used in the social share. | |
site.language | en | Language used by the generated HTML pages. |
site.title | Page title | Title used for the root index page. |
site.description | Page description | Description used for the root index page. |
theme.name | light | DaisyUI theme used by the generated website. |
content.ignorePatterns | [] | Glob patterns identifying files and directories that must not be copied or published. |
Ignore patterns are evaluated against paths relative to the bundle root. For example, README.md excludes only the root README, while drafts/** excludes the contents of the drafts directory. The source bundle is never modified: Kiso applies these exclusions while copying files to the destination directory, before loading, validating, and publishing the resulting bundle.
Publishing profiles
Publishing profiles allow the same OKF bundle to be published in different ways without changing its source content. Store profiles in .kiso/<profile-name>/configuration.yaml and select one with --profile:
kiso-cli build
kiso-cli build --profile public
kiso-cli build --profile internal
When no profile is selected, Kiso loads .kiso/configuration.yaml. When a profile is selected, Kiso loads only that profile's configuration; the default configuration is not read or merged.
Kiso MCP Server
kiso-mcp-server exposes the knowledge contained in an OKF bundle to applications compatible with the Model Context Protocol. It does not generate a website and does not modify the source bundle.
Download the MCP server for your operating system from the releases page, then start it with:
./kiso-mcp-server --source=examples/kb-acme-example-v0.2
The server loads the bundle and listens on port 8080 by default. Connect the MCP-compatible AI application to:
http://localhost:8080/mcp
The AI application can then search concepts, retrieve concept content, and use this knowledge as context when answering questions.
MCP tools
| Tool | Description |
|---|---|
search_concepts | Searches for concepts in the OKF bundle using a text query. |
get_concept_content | Returns the Markdown content of a concept from its concept ID. |
Options
| Option | Default | Description |
|---|---|---|
-s, --source | . | Directory containing the OKF bundle to serve. |
-h, --host | 0.0.0.0 | Host used by the MCP server. |
-p, --port | 8080 | Port used by the MCP server. |
For example, use another port with:
./kiso-mcp-server --source=examples/kb-acme-example-v0.2 --port=8081
Files in the repo
- .github
- applications
- documentation
- examples
- libraries
- website
- .gitignore
- AGENTS.md
- CLAUDE.md
- justfile
- LICENSE
- pom.xml
- README.md
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More tools
Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Turn any technical book PDF into a Claude Code skill — ready to study, reference, and use while you work.
AAS Core is the local, agent-first control plane for complete catalog discovery, agent-owned selection, stack validation, and planning, backed by 2,115+ agentic skills. Includes CLI, local MCP, catalog, plugins, and Workbench.
OfficeCLI is the first and best Office suite purpose-built for AI agents to read, edit, and automate Word, Excel, and PowerPoint files. Free, open-source, single binary, no Office installation required.