grok-import-mcp
# grok-import-mcp
MCP server for batch token management on [grok2api](https://github.com/hexonal/grok2api).
## Tools
| Tool | Description |
|------|-------------|
| `batch_add_tokens` | Batch add tokens to a specific pool |
| `batch_delete_tokens` | Batch delete tokens from all pools |
| `enable_nsfw` | Enable NSFW for specified or all tokens |
| `list_tokens` | List tokens in a pool or all pools |
## Environment Variables
| Variable | Description |
|----------|-------------|
| `GROK_API_URL` | grok2api server URL (e.g. `http://167.114.174.225:8000`) |
| `GROK_API_KEY` | API key for admin authentication (optional) |
## Claude Code Configuration
```json
{
"mcpServers": {
"grok-tokens": {
"command": "uvx",
"args": ["--from", "git+https://github.com/hexonal/grok-import-mcp.git", "grok-import-mcp"],
"env": {
"GROK_API_URL": "http://167.114.174.225:8000",
"GROK_API_KEY": ""
}
}
}
}
```
## Local Development
```bash
pip install -e .
GROK_API_URL=http://localhost:8000 python -m grok_mcp
```
TDQS
Scored across 4 tools
Each tool has a distinct verb: batch_add_tokens adds, batch_delete_tokens deletes, enable_nsfw toggles a flag, and list_tokens lists. The descriptions clarify scope (specific pool vs. all pools), so there is no overlap or ambiguity.
Three tools use a clear verb_noun pattern (list_tokens, enable_nsfw, batch_add_tokens), but the 'batch_' prefix on add/delete creates a slight inconsistency. Still, all names are snake_case and readable, with predictable structure.
Four tools is perfectly scoped for a focused token management utility. Each tool handles a distinct operation without unnecessary bloat or redundancy.
The surface is missing key lifecycle operations: delete only works across all pools (not a specific pool), there is no disable_nsfw counterpart, and no way to update token properties besides NSFW. These are significant gaps for a token management tool.