data-ai MCP Server
Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.
Discovered via github-topic:mcp and last synced 2mo ago.
Install instructions not detected yet
Check the source repository for the latest setup steps.
What you get
Metric
Minimum
3.10+
`pipx --version`
Command
`graphify codebuddy install`
`graphify codex install`
`graphify opencode install`
`graphify aider install`
`graphify claw install`
`graphify trae install`
`graphify hermes install`
`graphify amp install`
`graphify cursor install`
What it adds
PDF extraction
`.docx` and `.xlsx` support
Google Sheets rendering
Video/audio transcription (faster-whisper + yt-dlp)
MCP stdio server
Neo4j push support
FalkorDB push support
SVG graph export
Leiden community detection (Python < 3.13 only)
Ollama local inference
OpenAI / OpenAI-compatible APIs
Google Gemini API
Anthropic Claude API (`--backend claude`, uses `ANTHROPIC_API_KEY`)
AWS Bedrock (uses IAM, no API key)
Azure OpenAI Service (`--backend azure`, uses `AZURE_OPENAI_API_KEY` + `AZURE_OPENAI_ENDPOINT`)
SQL schema extraction
Live PostgreSQL introspection (`--postgres DSN`)
BYOND DreamMaker `.dm`/`.dme` AST extraction (may need a C compiler + `python3-dev` if no wheel matches your platform)
Terraform / HCL `.tf`/`.tfvars`/`.hcl` AST extraction
Pascal / Delphi `.pas`/`.dpr`/`.dpk`/`.inc` AST extraction (more accurate `calls`/`inherits` edges; falls back to a regex extractor when absent)
Chinese query segmentation (jieba)
No per-session state (for load-balanced / CI deployments)
DeepSeek backend
Raise output cap for dense corpora
When the log is enabled, also record full subgraph responses (off by default)
Everything above
Kimi Code backend
Per-call timeout in seconds for HTTP, claude-cli, and Anthropic SDK backends (default: 600)
Override the 512 MiB graph.json size cap — e.g. `700MB`, `2GB`, or plain bytes
Extensions
Reap idle stateful sessions after N seconds (`0` disables)
Ollama local inference URL
How many times to retry a rate-limited (429) request before giving up (default: 6; honors `Retry-After`)
Override LLM temperature for semantic extraction — e.g. `0.7`, or `none` to omit
`.md .mdx .qmd .html .txt .rst .yaml .yml` (markdown `[text](./other.md)` links and `[[wikilinks]]` become `references` edges between docs)
Used for
Ollama model name
Force graph rebuild even with fewer nodes
`.docx .xlsx` (requires `uv tool install graphifyy[office]`)
Claude (Anthropic) backend
Override Ollama KV-cache window size
Auto-enable Google Workspace export
`.pdf`
Anthropic-compatible endpoint URL (LiteLLM proxy, gateways, ...)
Minutes to keep Ollama model loaded
Backend for `graphify prs --triage`
`.png .jpg .webp .gif`
Model name for the Claude backend — for custom endpoints, use the model name/alias your server exposes
Azure OpenAI Service backend
Model override for triage
Default
OpenAI or OpenAI-compatible APIs
Azure resource endpoint URL
Set to `1` to turn on the local query log at `~/.cache/graphify-queries.log` (records each query/path/explain question + corpus path). Off by default — nothing is written unless you opt in (#1797)
Transport to serve on
OpenAI-compatible server URL (llama.cpp, vLLM, LM Studio, ...)
Azure API version override
Enable the query log and write it to this path instead of the default
HTTP bind port
Model name for the OpenAI backend — for self-hosted servers, use the model name/alias your server exposes (check its `/v1/models` endpoint), e.g. `LFM2.5-8B-A1B-UD-Q4_K_XL` for llama.cpp
AST parallelism thread count
Set to `1` to force the query log off (wins over the enable vars)