Setting Up a Project
First time with RaiSE?
Start with Getting Started — it covers the complete first-use flow in 5 steps. This page is a deep-dive into project setup options.
This guide covers advanced project setup — greenfield vs brownfield configuration, governance structure, and what goes where.
Greenfield: New Project¶
Initialize¶
This creates the .raise/ directory:
.raise/
├── manifest.yaml # Project metadata
├── config.toml # Project configuration
└── rai/
├── framework/
│ └── methodology.yaml # RaiSE methodology config
├── memory/
│ └── MEMORY.md # Memory index
└── personal/ # Your private data
└── .gitkeep
Set Up Governance¶
RaiSE creates a governance/ directory with templates:
governance/
├── constitution.md # Your principles
├── prd.md # Your requirements
├── guardrails.md # Your rules
├── backlog.md # Your work items
└── architecture/
├── system-context.md
└── system-design.md
Fill these in with your project's specifics:
-
Constitution — What are your non-negotiable principles? "Type annotations on all code." "Tests before implementation." Write 5-10 of these.
-
PRD — What are you building? Define 3-5 requirements at the feature level.
-
Guardrails — What rules must your code follow? Use MUST for non-negotiable, SHOULD for recommended. Link each guardrail to a requirement.
You don't need to fill everything at once. Start with a few principles and guardrails. Add more as you learn what matters for your project.
Set Up Skills¶
Skills live in .claude/skills/. RaiSE comes with a standard set of lifecycle skills. To see what's available:
You can create project-specific skills:
Build Memory¶
After filling governance files, build the memory index:
This creates the unified knowledge graph from all your governance documents, making them queryable and loadable into your AI's context.
After setup, continue with Getting Started — Step 5 to open your first session.
Brownfield: Existing Project¶
Initialize with Detection¶
For existing codebases, use --detect to analyze conventions:
This does everything rai init does, plus:
- Scans your source code for patterns
- Identifies coding conventions (naming, formatting, imports)
- Detects testing patterns and frameworks
- Generates guardrails from detected conventions
Review the generated governance/guardrails.md — it's a starting point, not gospel. Adjust to match your team's actual standards.
Discovery Scan¶
For deeper analysis, run the discovery pipeline:
# Scan source code
rai discover scan src/ --language python
# Analyze with confidence scoring
rai discover scan src/ -l python -o json | rai discover analyze
# Check for architectural drift later
rai discover drift
Discovery extracts your codebase's structure — classes, functions, modules — and builds a component map. This feeds into the knowledge graph, giving your AI partner architectural awareness.
Integrate with Existing Workflow¶
RaiSE adds structure alongside your existing tools:
- Git: RaiSE uses standard Git branching. Story branches are created from your development branch (
dev). Epics are logical containers (directories), not branches. - CI/CD: RaiSE doesn't touch your pipeline. Guardrails are enforced at the AI level, not the CI level.
- Editor: Skills are invoked through your AI assistant (e.g.,
/rai-story-startin Claude Code). No editor plugin needed.
Project Structure Reference¶
A fully set up RaiSE project looks like:
my-project/
├── .raise/ # RaiSE runtime
│ ├── manifest.yaml
│ └── rai/
│ ├── memory/ # Shared memory (committed)
│ ├── personal/ # Private data (gitignored)
│ └── identity/
├── .claude/
│ └── skills/ # Skill definitions
│ ├── session-start/
│ ├── story-start/
│ └── ...
├── governance/ # Project governance
│ ├── constitution.md
│ ├── prd.md
│ ├── guardrails.md
│ └── architecture/
├── work/ # Work tracking
│ └── epics/
│ └── e01-my-epic/
│ ├── SCOPE.md
│ └── stories/
├── src/ # Your code
└── tests/ # Your tests
What to Commit¶
| Directory | Commit? | Why |
|---|---|---|
.raise/rai/memory/ |
Yes | Shared patterns and calibration |
.raise/rai/personal/ |
No | Developer-specific, gitignored |
.raise/manifest.yaml |
Yes | Project metadata |
governance/ |
Yes | Shared governance documents |
.claude/skills/ |
Yes | Shared skill definitions |
work/epics/ |
Yes | Work tracking and retrospectives |