Skip to main content
Glama

CodeGraphContext

troubleshoot.md4.21 kB
# CodeGraphContext Troubleshooting Guide Use this checklist whenever `cgc setup` or `cgc start` doesn’t behave as expected. It keeps the happy path short, but includes the fallback steps when something goes wrong. ## 1. Prerequisites at a glance - **Windows + PowerShell** commands below assume the `py` launcher. Adapt to `python3` if you’re on macOS/Linux. - **Python 3.11** (recommended). Run `py -3.11 --version` to confirm. - **Git** for cloning the repository. - **Docker Desktop** installed *and running* before you launch `cgc setup` if you want the wizard to spin up a local Neo4j for you. - **Neo4j account** (only required if you prefer Neo4j AuraDB instead of Docker). > **Tip:** If Docker isn’t running, the setup wizard will fail when it tries to install Neo4j locally. ## 2. Create and activate a virtual environment From the repository root (`CodeGraphContext/`): ```powershell py -3.11 -m venv venv .\venv\Scripts\python.exe -m pip install --upgrade pip ``` - On Windows, Neo4j driver 6.x can crash with `AttributeError: socket.EAI_ADDRFAMILY`. If you see that, run: ```powershell .\venv\Scripts\python.exe -m pip install "neo4j<6" ``` ## 3. Run the setup wizard (preferred) Launch the wizard: ```powershell .\venv\Scripts\cgc.exe setup ``` What happens next: - The wizard checks for Docker. If it’s running, it can auto-provision a local Neo4j instance for you. - Alternatively, you can supply credentials for an existing Neo4j AuraDB database. - At the end, it generates: - `mcp.json` in your project directory (stores the MCP server command + env vars). - `~/.codegraphcontext/.env` containing `NEO4J_URI`, `NEO4J_USERNAME`, `NEO4J_PASSWORD`. Make sure the Docker container (or remote Neo4j) is still running before you start the server. ## 4. Start the MCP server Once the wizard completes successfully: ```powershell .\venv\Scripts\cgc.exe start ``` Expected output includes: ```text Starting CodeGraphContext Server... ... MCP Server is running. Waiting for requests... ``` If you instead see: ```text Configuration Error: Neo4j credentials must be set via environment variables ``` then either no credentials were saved, or the wizard was skipped—see the manual alternative below. ## 5. Manual credential setup (fallback) If you prefer not to use the wizard or need to fix a broken configuration: 1. Create a `mcp.json` (or edit the one that exists) in the repository root: ```json { "mcpServers": { "CodeGraphContext": { "command": "cgc", "args": ["start"], "env": { "NEO4J_URI": "neo4j+s://YOUR-HOSTNAME:7687", "NEO4J_USERNAME": "neo4j", "NEO4J_PASSWORD": "super-secret-password" } } } } ``` 2. (Optional) Also create `%USERPROFILE%\.codegraphcontext\.env` with the same key/value pairs. The CLI loads that file automatically. 3. Re-run: ```powershell .\venv\Scripts\cgc.exe start ``` ## 6. Common issues & fixes | Symptom | Likely Cause | Fix | | --- | --- | --- | | `Configuration Error: Neo4j credentials must be set…` | `mcp.json`/`.env` missing or empty | Run `cgc setup` again **with Docker running**, or create the files manually (section 5). | | `AttributeError: socket.EAI_ADDRFAMILY` | Neo4j 6.x bug on Windows | Install the 5.x driver: `.\venv\Scripts\python.exe -m pip install "neo4j<6"` and retry. | | Setup wizard fails while pulling Docker image | Docker Desktop not running or Docker permissions missing | Start Docker Desktop, wait for it to report “Running”, then rerun `cgc setup`. | | Server exits immediately with no log | Neo4j instance is offline | Check Docker container status or AuraDB dashboard; restart Neo4j and call `cgc start` again. | ## 7. After the server is running - Keep the virtual environment active whenever you run `cgc` commands. - Use `pytest` from the same env to run tests: ```powershell .\venv\Scripts\pytest ``` - Front-end website lives under `website/` if you need to run `npm run dev`. When in doubt, re-run the wizard with Docker active—it regenerates the configuration files without touching your code. Let me know if any section needs clarifying! :)

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Shashankss1205/CodeGraphContext'

If you have feedback or need assistance with the MCP directory API, please join our Discord server