How to Share the Egonex AI Knowledge Graph JSON with Team Members for Onboarding

Copy the knowledge-graph.json file from the .understand-anything/ directory to a shared location or GitHub release asset after running the analysis command, allowing new team members to explore the complete codebase structure instantly without waiting for a full analysis run.

The Egonex AI Understand-Anything tool generates a self-contained knowledge graph JSON that serves as a portable architectural snapshot of your repository. This file encodes file relationships, language tags, and symbol dependencies in a lightweight format that lives at .understand-anything/knowledge-graph.json. Sharing this JSON streamlines onboarding by eliminating the setup time typically required for initial repository analysis.

Where the Knowledge Graph JSON Is Generated

Graph Construction Pipeline

The core analyzer located at packages/core/src/analyzer/graph-builder.ts traverses the project and parses source files using the built-in tree-sitter parser. It assembles a directed graph of nodes representing files, symbols, frameworks, and their relationships, creating a comprehensive structural map of the codebase.

Persistence Layer

Once the graph is built, the persistence module writes the result to the canonical file name defined by the constant GRAPH_FILE = "knowledge-graph.json" inside the hidden .understand-anything directory. This logic is implemented in packages/core/src/persistence/index.ts.

Development Server Endpoint

When the dashboard runs locally, the Vite development server configured in packages/dashboard/vite.config.ts includes middleware that serves the file at the route /knowledge-graph.json. This endpoint is accessible to the web UI and can be queried directly using standard HTTP requests or curl commands.

Generating the JSON File for Distribution

To create a shareable snapshot of your codebase, execute the build and analysis commands from the project root:

# Build the core analysis package

pnpm --filter @understand-anything/core build

# Run the full analysis to generate the knowledge graph

pnpm run understand --full

After completion, the JSON file will be available at:

.understand-anything/knowledge-graph.json

Methods for Sharing the JSON with Your Team

Direct File Distribution

Copy the generated file to a shared location accessible to your team members:

# Create a dedicated onboarding directory

mkdir -p ./onboarding

# Copy the knowledge graph for distribution

cp .understand-anything/knowledge-graph.json ./onboarding/knowledge-graph.json

You can then place this copy on a shared drive, in a team wiki, or any central documentation repository.

GitHub Release Assets

Attach the JSON to a GitHub release to version it alongside your code:

1. Navigate to your repository's releases page
2. Create a new release or draft
3. Upload knowledge-graph.json as a release asset
4. Tag the release with the corresponding git commit SHA for traceability

Repository Storage

Store the file within the repository itself for version control:

• Create a folder such as onboarding/ at the repository root
• Copy the JSON file into this directory
• Add .understand-anything/ to .gitignore to prevent developers from accidentally committing their local analysis runs while keeping the onboarding/ folder tracked

Loading the Shared JSON for Onboarding

New team members can start exploring the codebase immediately using the shared JSON without re-running the analysis:

1. Copy the shared knowledge-graph.json into their local .understand-anything/ directory
2. Start the development dashboard:

pnpm dev:dashboard

3. Open http://localhost:5173 in a browser. The dashboard automatically fetches the graph from /knowledge-graph.json and renders the interactive visualization.

Best Practices for Maintaining Shared Graphs

• Regenerate after major changes: Run pnpm run understand --full following significant refactors, new module additions, or dependency updates to ensure the graph reflects the current architecture.

• Verify no secrets are embedded: While the graph contains only structural metadata, audit the JSON before sharing to confirm that no source files containing sensitive credentials were inadvertently parsed into node attributes.

• Use clean copies: Always copy the file out of .understand-anything/ before distributing to prevent accidental overwrites from other developers' local analysis runs.

• Document the version: Include the git commit SHA or version tag that produced the snapshot, either in a header comment within the JSON or in an accompanying README.md file.

Summary

• The knowledge graph JSON is generated at .understand-anything/knowledge-graph.json by the graph-builder.ts analyzer and persisted via packages/core/src/persistence/index.ts.
• Run pnpm run understand --full to generate the file, then copy it to a shared location or GitHub release asset.
• The Vite dev server in packages/dashboard/vite.config.ts serves the JSON at /knowledge-graph.json for immediate dashboard visualization.
• New team members load the shared JSON by placing it in their local .understand-anything/ directory and running pnpm dev:dashboard.

Frequently Asked Questions

Where is the Egonex AI knowledge graph JSON stored?

The file is stored at .understand-anything/knowledge-graph.json relative to the project root. This location is defined by the GRAPH_FILE constant in packages/core/src/persistence/index.ts.

Can I share the knowledge graph JSON without sharing the entire repository?

Yes. The JSON file is self-contained and includes all structural information necessary to visualize the codebase architecture. Recipients can view the graph in the dashboard by placing the file in their own .understand-anything/ directory, though they will need the Understand-Anything tool installed to run the visualization interface.

How often should I update the shared knowledge graph JSON?

Update the shared JSON whenever the codebase undergoes significant structural changes, such as major refactors, addition of new modules, or framework migrations. For active development, consider regenerating the graph weekly or before each major release to ensure onboarding materials remain accurate.

Is the JSON file safe to share publicly?

Generally yes, as the graph contains only structural metadata (file paths, language tags, symbol relationships) rather than source code content. However, review the file before distribution to ensure that node attributes do not inadvertently contain sensitive strings from parsed files, and confirm that your organization's security policies permit sharing architectural metadata.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}