Chub

The missing context layer for AI-assisted development teams.

Built on Context Hub by Andrew Ng — Chub is a high-performance Rust rewrite that extends the original with team-first features: shared doc pinning, git-tracked annotations, context profiles, and agent config sync.

For individuals: drop-in replacement for chub with a faster binary, better search, and persistent annotations. For teams: commit a .chub/ directory to your repo and every developer and every AI agent gets the same versioned context, automatically.

What Chub does

Coding agents hallucinate APIs and forget what they learn between sessions. Context Hub's answer — curated, versioned markdown docs served via CLI and MCP — works well. Chub keeps that foundation and adds:


Native speed	27x faster builds, 26ms cold start, 1.2 MB binary — no Node.js required
Team pins	Lock docs to specific versions so every agent on the team uses the same reference
Shared annotations	Team knowledge lives in `.chub/annotations/` — committed to git, surfaced automatically
Custom project context	Your architecture docs, API conventions, and runbooks, served alongside public docs
Context profiles	Scope which docs and rules each role (backend, frontend, etc.) gets
Agent config sync	Generate and keep `CLAUDE.md`, `.cursorrules`, `AGENTS.md` in sync from one source
Private registry	Self-host internal docs alongside the public registry — no cloud required
Full compatibility	Same registry format, search index, and config schema as Context Hub

Installation

npm install -g @nrl-ai/chub

Or via pip:

pip install chub

Or download a pre-built binary from GitHub Releases.

Usage

# Search and fetch docs
chub search "stripe"
chub get openai/chat --lang python
chub list

# Initialize project for team sharing
chub init
chub init --from-deps          # auto-detect dependencies

# Pin docs for the team
chub pin openai/chat --lang python --version 4.0
chub pins                      # list pins
chub get --pinned              # fetch all pinned docs

# Context profiles
chub profile use backend
chub profile list

# Team annotations
chub annotate openai/chat "Use streaming API" --team

# Auto-detect dependencies and pin matching docs
chub detect --pin

# Generate CLAUDE.md, .cursorrules from .chub/config.yaml
chub agent-config generate

# Doc freshness check
chub check --fix

# Snapshots
chub snapshot create v1.0
chub snapshot diff v1.0 v2.0

# MCP server
chub mcp

# JSON output (all commands)
chub search "stripe" --json

MCP Integration

Add to your MCP config (.mcp.json for Claude Code, .cursor/mcp.json for Cursor):

{
  "mcpServers": {
    "chub": {
      "command": "chub",
      "args": ["mcp"]
    }
  }
}

MCP tools: chub_search, chub_get, chub_list, chub_annotate, chub_feedback. Registry resource: chub://registry.

Team Features

The .chub/ directory at the project root is committed to git and shared with the team.

my-project/
├── .chub/
│   ├── config.yaml          # Project-level config
│   ├── pins.yaml            # Pinned doc versions
│   ├── annotations/         # Team knowledge, git-tracked
│   ├── context/             # Custom docs: architecture, conventions
│   └── profiles/            # Role-scoped context (backend.yaml, frontend.yaml)

Three-tier config inheritance

~/.chub/config.yaml          # Tier 1 — personal defaults
    ↓ overridden by
.chub/config.yaml            # Tier 2 — project config (shared)
    ↓ overridden by
.chub/profiles/<name>.yaml   # Tier 3 — role/task profile

Implemented features

Feature	Status	Description
Project init (`chub init`)	Done	Create `.chub/` with sensible defaults
Doc pinning	Done	Lock doc versions in `pins.yaml`
Team annotations	Done	Git-tracked annotations with merge
Context profiles	Done	Role-scoped context with inheritance
Project context	Done	Custom markdown docs served via MCP
Dep auto-detection	Done	9 file types (npm, Cargo, pip, Go, etc.)
Agent config sync	Done	Generate CLAUDE.md, .cursorrules, etc.
Doc snapshots	Done	Point-in-time pin snapshots
Doc freshness	Done	Compare pinned vs installed versions
Usage analytics	Done	Local opt-in fetch tracking

See docs/plan.md for the full roadmap.

Benchmarks

Measured on the production corpus (1,553 docs, 6 skills, 1,691 files).

Build

Operation	JS (`chub`)	Rust (`Chub`)	Speedup
`build` (4 entries)	1,050 ms	38 ms	27x
`build` (1,559 entries)	6,300 ms	2,500 ms	2.5x
`build --validate-only`	6,300 ms	360 ms	17x
Cold start (`--help`)	120 ms	26 ms	4.6x

Resource Usage

Metric	JS	Rust
Binary size	~70 MB (with `node_modules`)	1.2 MB
Runtime dependency	Node.js 20+	None (single binary)
Memory (build, 1,559 entries)	~120 MB	~15 MB

Test Suite

99 tests covering behavioral parity and team features:

Suite	Tests	Coverage
Tokenizer	6	Stop words, punctuation, edge cases
BM25 search	7	Scoring, ranking, limits, field weights
Inverted index	1	Parity with linear scan
Frontmatter parser	9	YAML, CRLF, BOM, empty, numeric
Language normalization	4	Aliases, case, unknown
Build integration	15	Output format, validation, structure
Search parity	20	Multi-word, tags, descriptions
Team features	33	Pins, profiles, annotations, snapshots, detect, freshness, agent config, analytics

All tests use isolated temp directories — no writes to the repo's .chub/.

Documentation

Full documentation at chub.nrl.ai (VitePress):

License

MIT

chub-cli 0.1.1