Contributing to Lucid Layer

Thank you for your interest in contributing to Lucid — the autonomous AI infrastructure layer. Whether you’re fixing a bug, adding a feature, improving docs, or writing tests, your contribution matters.

Getting Started
Development Workflow
Project Structure
Code Standards
Commit Conventions
Testing
Working with Solana Programs
Working with EVM Contracts
Working with the Offchain Engine
Pull Request Process
Reporting Issues
Architecture Guidelines
License

Getting Started

# 1. Fork and clone
git clone https://github.com/<your-username>/Lucid-L2.git
cd Lucid-L2

# 2. Install offchain dependencies
cd offchain && npm install

# 3. Copy environment config
cp .env.example .env

# 4. Verify everything works
npm test                  # 103 suites, 1,683 tests
npm run type-check        # TypeScript compilation

Prerequisites

Tool	Version	Required for
Node.js	20+	Offchain engine + API
npm	9+	Package management
Rust + Cargo	Latest stable	Solana programs
Solana CLI	1.18+	On-chain development
Anchor	0.30+	Solana program framework
Docker	24+	Local deployment testing

Development Workflow

# 1. Create a feature branch from master
git checkout -b feat/your-feature

# 2. Make changes

# 3. Run all checks
cd offchain
npm run type-check        # Must pass — no new type errors
npm test                  # Must pass — no regressions
npm run lint              # Must pass — no lint violations

# 4. Commit (see conventions below)
git commit -m "feat(memory): add temporal decay to semantic recall"

# 5. Push and open a PR
git push origin feat/your-feature

Project Structure

Lucid-L2/
├── programs/              6 Solana Anchor programs (Rust)
├── contracts/             17 EVM contracts (Solidity)
├── offchain/
│   └── packages/
│       ├── engine/        Truth library — no HTTP, no Express
│       │   └── src/
│       │       ├── identity/       Passports, wallets, NFT, shares
│       │       ├── memory/         6 memory types, recall, compaction
│       │       ├── epoch/          Epoch lifecycle, MMR
│       │       ├── receipt/        Cryptographic receipts
│       │       ├── payment/        x402, pricing, splits, escrow
│       │       ├── compute/        Deployers, runtime adapters, agents
│       │       ├── deployment/     Control plane (state, reconciler, rollout)
│       │       ├── anchoring/      Unified DePIN dispatcher
│       │       ├── reputation/     On/off-chain reputation
│       │       └── shared/         Crypto, DB, config, chains, jobs
│       ├── gateway-lite/  Express API — thin route handlers
│       └── sdk/           Developer SDK
├── schemas/               14 JSON validation schemas
├── tests/                 Solana program tests (Mocha)
├── examples/              Quickstart guides (JS, Python, TS)
└── docs/                  Design specs + implementation plans

Critical dependency rule:

gateway-lite may import from engine
engine must NEVER import from gateway-lite (ESLint-enforced)

Code Standards

TypeScript

Strict mode — all new code must compile under strict: true
No any — use proper types. unknown + type guards when truly dynamic.
No import type in test files — babel/jest doesn’t support it. Use regular import.
Feature-domain organization — code lives in its domain folder (memory, payment, deployment), not grouped by technical layer
Interface-first — new features should define an I* interface, then implement (e.g., IDeploymentStore → PostgresDeploymentStore + InMemoryDeploymentStore)
Factory pattern — use getXxx() singletons with resetXxx() for test teardown

Naming

Element	Convention	Example
Files	camelCase	`agentDeploymentService.ts`
Interfaces	`I` prefix	`IDeploymentStore`
Types/Enums	PascalCase	`ActualState`, `DeploymentEventType`
Constants	UPPER_SNAKE	`VALID_TRANSITIONS`
Functions	camelCase	`getDeploymentStore()`
Test files	`*.test.ts`	`control-plane.test.ts`

What NOT to do

Don’t add comments to code you didn’t change
Don’t add error handling for scenarios that can’t happen
Don’t create abstractions for one-time operations
Don’t add backward-compatibility shims — just change the code
Don’t introduce circular dependencies between packages

Commit Conventions

Follow Conventional Commits:

<type>(<scope>): <description>

[optional body]

Co-Authored-By: Your Name <your@email.com>

Types

Type	When
`feat`	New feature
`fix`	Bug fix
`refactor`	Code restructuring (no behavior change)
`test`	Adding or updating tests
`docs`	Documentation only
`chore`	Build, CI, tooling, dependencies

Scopes

Use the domain: memory, deployment, payment, receipt, epoch, identity, anchoring, reputation, compute, db.

Examples

feat(deployment): add blue-green rollout with slot promotion
fix(memory): prevent hash chain break on concurrent writes
refactor(anchoring): unify DePIN dispatch into single interface
test(deployment): 15 rollout tests — promote, rollback, cancel
docs: update README with autonomous stack overview

Testing

Running tests

cd offchain

# Full suite
npm test                                          # 103 suites, 1,683 tests

# Single file
npx jest packages/engine/src/deployment/__tests__/rollout.test.ts --no-coverage

# Pattern match
npx jest --testPathPattern="memory" --no-coverage

# Type check only
npm run type-check

Writing tests

Every new feature needs tests. No exceptions.
Use InMemory* stores for unit tests (fast, deterministic)
Mock external dependencies — deployers, providers, LLMs
Follow existing patterns — look at nearby __tests__/ folders
Test the contract, not the implementation — test via the interface, not internal methods

// Good: test through the interface
const store = new InMemoryDeploymentStore();
const deployment = await store.create({ agent_passport_id: 'agent-001', ... });
expect(deployment.actual_state).toBe('pending');

// Bad: test internal Map directly
expect(store['deployments'].size).toBe(1);

Test structure

engine/src/<domain>/__tests__/<domain>.test.ts

Each domain has its own __tests__/ folder next to the source code.

Working with Solana Programs

# Build all 6 programs
anchor build

# Run tests (requires local validator)
solana-test-validator --reset --quiet &
cd tests && npm test

# Deploy to devnet
anchor deploy --provider.cluster devnet

# Test specific program
cd tests && npm run test:passports

Program layout

Each program lives in programs/<name>/src/lib.rs with Anchor macros. PDAs follow the pattern ["seed", key.as_ref()].

Working with EVM Contracts

cd contracts
npx hardhat test
npx hardhat deploy --network base    # Deploy to Base

17 contracts in contracts/src/ — full parity with Solana programs where applicable.

Working with the Offchain Engine

Adding a new feature

Define the interface in engine/src/<domain>/
Implement InMemory (tests) + Postgres (production) versions
Add factory in index.ts with getXxx() / resetXxx()
Write tests in __tests__/ using the InMemory implementation
Add routes in gateway-lite/src/routes/ (thin handlers, delegate to engine)
Update OpenAPI in openapi.yaml

Key patterns to follow

Pattern	Where to look
Interface + InMemory + Postgres	`deployment/control-plane/store.ts`
Factory + singleton + reset	`deployment/control-plane/index.ts`
Status machine + transitions	`deployment/control-plane/state-machine.ts`
Optimistic locking	`deployment/control-plane/in-memory-store.ts`
DePIN dispatch	`anchoring/dispatcher.ts`
Memory type managers	`memory/managers/`
Route handlers (thin)	`gateway-lite/src/routes/`

Pull Request Process

Branch from master — use feat/, fix/, refactor/ prefixes
All checks must pass — type-check + tests + lint
No regressions — test count should stay the same or increase
One concern per PR — don’t mix features with refactors
Describe the why — PR description should explain motivation, not just list changes
Update docs if needed — CLAUDE.md, OpenAPI, README

PR template

## Summary
- What this PR does and why

## Changes
- List of key changes

## Test plan
- [ ] New tests added
- [ ] Existing tests pass
- [ ] Type check passes

Reporting Issues

Open a GitHub issue with:

What you expected to happen
What actually happened (include error messages, logs)
Steps to reproduce (minimal, specific)
Environment — OS, Node.js version, Solana CLI version, relevant env vars

For security vulnerabilities, email security@raijinlabs.io instead of opening a public issue.

Architecture Guidelines

Before making significant changes, understand the core principles:

Lucid is the control plane, not the execution authority. Agents run on decentralized providers. Lucid coordinates — it doesn’t own.
Local truth, global supervision. Agent memory is agent-owned (SQLite). Deployment state is fleet-wide (Supabase). Never mix these.
Interface-first, swap later. Every external dependency (IDepinStorage, INFTProvider, IDeploymentStore, ISecretsResolver) is behind a swappable interface.
Events, not coupling. State changes emit events. Consumers react. Don’t create direct cross-domain function calls.
L3 is operational, not canonical. Supabase stores operational projections. Chain + DePIN are the source of truth. If Supabase is lost, rebuild from L1+L2.

Read CLAUDE.md for the full architecture reference.

License

By contributing, you agree that your contributions will be licensed under:

Engine & programs: Apache License 2.0
Gateway-lite: AGPL-3.0

Self-Hosting

Advanced

Contributing

Contributing to Lucid Layer

Table of Contents

Getting Started

Prerequisites

Development Workflow

Project Structure

Code Standards

TypeScript

Naming

What NOT to do

Commit Conventions

Types

Scopes

Examples

Testing

Running tests

Writing tests

Test structure

Working with Solana Programs

Program layout

Working with EVM Contracts

Working with the Offchain Engine

Adding a new feature

Key patterns to follow

Pull Request Process

PR template

Reporting Issues

Architecture Guidelines

License

Self-Hosting

Advanced

Documentation Index

​Contributing to Lucid Layer

​Table of Contents

​Getting Started

​Prerequisites

​Development Workflow

​Project Structure

​Code Standards

​TypeScript

​Naming

​What NOT to do

​Commit Conventions

​Types

​Scopes

​Examples

​Testing

​Running tests

​Writing tests

​Test structure

​Working with Solana Programs

​Program layout

​Working with EVM Contracts

​Working with the Offchain Engine

​Adding a new feature

​Key patterns to follow

​Pull Request Process

​PR template

​Reporting Issues

​Architecture Guidelines

​License

Contributing to Lucid Layer

Table of Contents

Getting Started

Prerequisites

Development Workflow

Project Structure

Code Standards

TypeScript

Naming

What NOT to do

Commit Conventions

Types

Scopes

Examples

Testing

Running tests

Writing tests

Test structure

Working with Solana Programs

Program layout

Working with EVM Contracts

Working with the Offchain Engine

Adding a new feature

Key patterns to follow

Pull Request Process

PR template

Reporting Issues

Architecture Guidelines

License