Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
78 lines (58 loc) · 2.39 KB

CONTRIBUTING.md

File metadata and controls

78 lines (58 loc) · 2.39 KB

Contributing to Yaad

Thank you for your interest in contributing! Yaad is a memory layer for coding agents — your contributions help make it better for everyone.

Quick Start

git clone https://github.com/GrayCodeAI/yaad
cd yaad
make build    # verify it builds
make test     # run tests

Requirements: Go 1.25+. No CGO, no C compiler needed.

What to Work On

Check open issues for things to pick up.

Good first issues:

  • Improve entity extraction patterns in internal/engine/entities.go
  • Add a new memory node type
  • Improve privacy filter patterns in internal/privacy/filter.go
  • Add export formats in internal/exportimport/export.go

Development

# Build
make build

# Test
make test

# Lint
go vet ./...

Pull Request Guidelines

  1. One thing per PR — keep it focused
  2. Tests required — add a test for new functionality
  3. No CGO — Yaad must build with CGO_ENABLED=0
  4. No LLM API calls in hot paths — Yaad is a memory layer, not an LLM client
  5. Keep it minimal — avoid unnecessary abstractions
  6. Localhost-only — REST server must never bind to public interfaces

Architecture

See ARCHITECTURE.md for the full technical design.

Key principles:

  • Yaad is a memory layer. It stores, retrieves, and organizes memories.
  • MCP-first. The primary integration is via MCP stdio — agents call yaad mcp.
  • Single-user. No auth, no multi-tenancy. One developer, one machine.
  • Pure Go. No CGO, no external dependencies at runtime.

Project Structure

cmd/yaad/           CLI entry point (cobra commands)
internal/
  engine/           Core memory engine (remember, recall, context, decay)
  graph/            DAG operations (BFS, impact, ancestors, subgraph)
  storage/          SQLite storage layer (FTS5, WAL mode)
  server/           REST API + MCP server
  hooks/            Auto-capture hooks (session lifecycle)
  compact/          Memory compaction (summarize old memories)
  config/           TOML config loading
  privacy/          Secret detection and redaction
  skill/            Procedural memory (reusable step sequences)
  git/              Git-aware staleness detection
  embeddings/       Vector embedding providers (OpenAI, Voyage, local stub)

License

By contributing, you agree your contributions are licensed under MIT.