Contributing

Contributing

We welcome contributions to ATP! Here's how to get involved.

---

Getting Set Up

git clone https://github.com/rajamohan1950/AgentTransportProtocol.git
cd AgentTransportProtocol

export PROTOC=$(which protoc)

cargo build --workspace
cargo test --workspace     # All 280 tests must pass
cargo clippy --workspace   # Zero warnings required

Areas of Interest

High Priority

Area	Crate	Description
gRPC Server	atp-transport	Flesh out the full gRPC server with all 10 RPCs
gRPC Client	atp-transport	Client-side implementation for networked agents
Integration Tests	(new)	Cross-layer end-to-end test scenarios
Python SDK	atp-python	Expand PyO3 bindings for all layer functions

Medium Priority

Area	Crate	Description
Real-world benchmarks	atp-bench	Benchmarks with actual LLM API calls
Persistent storage	atp-identity	Database-backed identity and trust store
WebSocket transport	atp-transport	Alternative to gRPC for browser environments
Observability	(new)	Metrics, tracing, and logging across layers

Research

Area	Description
Embedding models	Replace hash-based embeddings with real models in L3
Adaptive routing	Reinforcement learning for route pattern selection
Federated trust	Cross-network trust propagation
Formal verification	Prove correctness of handshake and routing protocols

Code Style

• Rust edition: 2021
• Formatting: cargo fmt
• Linting: cargo clippy with zero warnings
• Tests: Every public function must have tests
• Documentation: Doc comments on all public items

Pull Request Process

1. Fork the repository
2. Create a feature branch: git checkout -b feat/my-feature
3. Write your code and tests
4. Ensure all checks pass:

   cargo fmt --check
   cargo clippy --workspace
   cargo test --workspace

5. Open a PR with a clear description of what and why

Project Structure

crates/
├── atp-types/      # Core types — change carefully, everything depends on this
├── atp-proto/      # Generated code — modify proto/ files, not the generated code
├── atp-identity/   # Layer 1 — trust formulas here
├── atp-handshake/  # Layer 2 — handshake state machine here
├── atp-context/    # Layer 3 — compression algorithms here
├── atp-routing/    # Layer 4 — Bellman-Ford and patterns here
├── atp-fault/      # Layer 5 — circuit breaker logic here
├── atp-transport/  # gRPC stubs — NEEDS WORK
├── atp-node/       # Composition root — NEEDS WORK
├── atp-sim/        # Simulation — add new agent behaviors here
├── atp-bench/      # Benchmarks — add new scenarios here
├── atp-sdk/        # Public facade — keep it simple
└── atp-python/     # Python bindings — built separately with maturin

Questions?

Open an issue on GitHub or reach out to Rajamohan Jabbala.