Skip to main content
Glama
thecodacus

OKF Knowledge Agent MCP Server

by thecodacus

OKF Knowledge Agent

An LLM-managed knowledge base following the Open Knowledge Format (OKF) v0.1 spec — a bundle of plain markdown files with YAML frontmatter, readable by humans, diffable in git, managed by an agent.

Three ways in, one agent:

  • MCP serverkb_query / kb_add / kb_update / kb_status tools over stdio or streamable HTTP. Each call drives an internal LLM agent with the OKF spec in its system prompt.

  • Web UI — browse the bundle (tree, concept viewer, update log, conformance badge) and chat with the same agent to test it. Tool calls render inline so you can watch it work.

  • CLIpnpm agent:query "..." / pnpm agent:mutate "..." smoke entries.

Design rule: conformance is enforced in code, not prompts. The deterministic bundle layer validates frontmatter (type required), regenerates index.md files, appends log.md entries (newest-first, spec §7), and sandboxes all paths to the bundle root. The LLM decides what to change; the code guarantees the result is a conformant bundle.

Stack

pnpm monorepo:

Package

What

packages/core

OKF bundle layer (zero LLM) + agent (Vercel AI SDK tool loop: search/read/list/write/patch/delete) + provider registry

packages/server

Fastify: MCP streamable-HTTP at /mcp, stdio bin, REST browse API at /api/*, streaming chat at /api/chat, serves the web build

packages/web

Vite + React + TS + Tailwind: bundle browser + agent chat (useChat)

Providers (env-selected, swappable per chat): Anthropic (default), OpenRouter, llamacpp (llama.cpp llama-server / llama-swap — model auto-discovered from /v1/models, loaded model preferred), local (any other OpenAI-compatible endpoint).

llama.cpp

# on the inference box — --jinja enables OpenAI-style tool calling
llama-server -m model.gguf --jinja --host 0.0.0.0 --port 8080

# here — no model id needed, it's discovered
LLM_PROVIDER=llamacpp LLAMACPP_BASE_URL=http://inference-box:8080 \
BUNDLE_ROOT=./sample-bundle node packages/server/dist/index.js

Works behind llama-swap too: discovery prefers the currently loaded model so a query doesn't trigger a multi-minute model swap. Pin a specific model with LLM_MODEL=.

Related MCP server: Kremis

Quick start

pnpm install
pnpm build
cp .env.example .env   # add your API key

BUNDLE_ROOT=./sample-bundle ANTHROPIC_API_KEY=sk-... node packages/server/dist/index.js
# → http://localhost:3800  (web UI + /api + /mcp)

Dev mode (server on :3800, Vite HMR on :5180 with proxy):

BUNDLE_ROOT=./sample-bundle pnpm --filter @okf-agent/server dev
pnpm --filter @okf-agent/web dev

MCP registration (Claude Code / Desktop)

claude mcp add okf-kb \
  -e BUNDLE_ROOT=/path/to/your/bundle \
  -e ANTHROPIC_API_KEY=sk-... \
  -- node /path/to/okf-agent/packages/server/dist/mcp/stdio.js

Or point an HTTP MCP client at http://host:3800/mcp.

Docker

docker compose up --build
# bundle is a volume mount — point ./sample-bundle at any OKF bundle

Tests

pnpm test                                  # core: 15 tests (spec §5/§6/§7/§9, sandbox, search, concurrency)
pnpm --filter @okf-agent/server exec tsx scripts/mcp-smoke.mts   # MCP stdio round-trip (needs SMOKE_BUNDLE + an API key)

Environment

See .env.example. BUNDLE_ROOT is required; GIT_AUTOCOMMIT=true commits every mutation.

F
license - not found
-
quality - not tested
C
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/thecodacus/okf-agent'

If you have feedback or need assistance with the MCP directory API, please join our Discord server