Graphify Docs

Appearance

Solo Developer Workflow ​

Fresh

Overview ​

The solo developer workflow covers the typical pattern for individual use: initial build, incremental updates, and querying during development.

Initial Setup ​

flowchart LR
    A[Clone project] --> B[pip install graphifyy]
    B --> C[graphify install]
    C --> D[Run graphify]
    D --> E[Graph built]
    E --> F[Start querying]

Step-by-Step ​

1. Install Graphify ​

bash
pip install graphifyy
graphify install   # or: graphify claude install

2. Build the Initial Graph ​

bash
cd /path/to/your/project
/graphify .

On a large codebase, this takes a few minutes. Go make coffee.

3. Review the Report ​

bash
cat graphify-out/GRAPH_REPORT.md

Read the God Nodes and community clusters. This gives you a mental model of the codebase before reading a single file.

4. Query Before Reading ​

Instead of opening random files, query first:

bash
/graphify query "payment processing"
/graphify explain "PaymentService"
/graphify path "APIController" "Database"

5. Update After Changes ​

After writing code:

bash
/graphify . --update

Takes seconds on large codebases because only changed files are reprocessed.

Daily Workflow Loop ​

flowchart LR
    A[Start work session] --> B[graphify update]
    B --> C[Query graph for context]
    C --> D[Code changes]
    D --> E{Significant changes?}
    E -->|Yes| F[graphify update]
    E -->|No| G[Continue coding]
    F --> C
    G --> D

Keeping the Graph Committed ​

Add graphify-out/ to your repository so the graph is available immediately when you clone or switch branches:

bash
git add graphify-out/graph.html graphify-out/graph.json graphify-out/GRAPH_REPORT.md
git add -f graphify-out/wiki/      # if you're using --wiki export
git commit -m "Update knowledge graph"

Add to .gitignore:

gitignore
graphify-out/cache/    # Each machine rebuilds cache locally

Tips ​

First Query on a New Codebase

Start with GRAPH_REPORT.md. The God Nodes tell you the most important concepts. Read those files first — they're the mental model of the whole system.

Update Frequency

The --update flag is cheap. Run it after every feature branch, before debugging sessions, and when switching between tasks.