v0.1.0 — First release

kb — Headless Knowledge Base Engine

Single Go binary that turns codebases into searchable, LLM-compiled knowledge wikis. No embeddings, no vectors.

Features

• Build — Scan files, parse AST (Go/Python/TypeScript), compile with LLM
• Search — BM25 with title + concept boosting, ~10ms latency
• Cache — SHA256 content hashing, instant warm builds
• Parallel — 5 concurrent LLM compilations
• Watch — Auto-rebuild on file changes
• Lint — Structural + LLM-powered health checks
• Export — --output flag for readable markdown wikis

Install

# Download binary for your platform
# macOS ARM (M1/M2/M3):
curl -L https://github.com/qbtrix/kb-go/releases/download/v0.1.0/kb-darwin-arm64 -o kb && chmod +x kb

# Or build from source:
go install github.com/qbtrix/kb-go@v0.1.0

Quick Start

export ANTHROPIC_API_KEY="sk-..."
kb build ./src --scope myapp --pattern "*.go,*.py,*.ts"
kb search "auth middleware" --scope myapp

Benchmarks (claude-haiku-4-5-20251001)

Codebase	Files	Cold Build	Warm Build	Search Relevance
litestream (Go)	129	355s	0.03s	—
flask (Python)	83	78s	0.01s	—
small corpus	10	9.5s	0.01s	100%