Graphify

Graphify turns a project into a queryable knowledge graph instead of leaving an AI coding assistant to grep and re-read files on every question. Running /graphify inside a compatible tool maps code, documentation, PDFs, images, and even videos into a unified graph, and the resulting graphify-out/ output is meant to be committed to git so the whole team starts from the same shared map rather than each person’s assistant re-indexing independently.

It works across Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and other AI coding assistants as a skill rather than a tool tied to one vendor, and can run in MCP server mode (/graphify ./raw --mcp) for direct agent integration. Beyond application code, it explicitly covers SQL schemas, R scripts, shell scripts, and infrastructure definitions in the same graph, aiming to represent a whole system rather than just the app layer.

MIT licensed and backed by Y Combinator (S26 batch), Graphify is distributed via PyPI (graphifyy) with public download tracking, and localized into more than 25 languages — reflecting substantial investment behind what started as an individual developer’s project before growing into a company.

What You Get

• A /graphify command that maps an entire project into a knowledge graph across Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and more
• Coverage beyond application code: SQL schemas, R scripts, shell scripts, docs, PDFs, images, and videos in one graph
• A shareable graphify-out/ output meant to be committed to git so a whole team starts from the same map
• MCP server mode for direct integration into agent tool-calling workflows

Common Use Cases

• Giving an AI coding assistant structured awareness of app code, database schema, and infrastructure together instead of separately
• Sharing one team-wide project knowledge graph via git instead of every developer’s assistant re-indexing independently
• Querying relationships across code, docs, and even non-code assets like PDFs or diagrams in one unified structure
• Integrating project-graph context into an agent’s tool-calling loop via the MCP server mode

Under The Hood

Architecture Graphify processes a project’s files — source code, SQL schemas, scripts, documents, and media — into a unified graph representation stored in a graphify-out/ directory designed to be committed alongside the code itself, turning the graph into a versioned artifact rather than an ephemeral local index. Its MCP server mode exposes this graph directly to agent tool-calling loops, meaning agents don’t need a separate retrieval step outside their normal MCP tool usage.

Tech Stack Python, distributed via PyPI as graphifyy, working as a cross-tool skill compatible with Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and others rather than a single-vendor plugin. CI is run via GitHub Actions, and the project maintains translated documentation in more than 25 languages.

Code Quality The project has grown from an individual developer’s repository into a Y Combinator-backed company (YC S26) with real download tracking via pypi.tech, a public CI badge, and multiple contributors with substantial commit counts — despite initial skepticism warranted by its unusually fast star growth, forks (7,600+), issue activity (400+), and contributor spread are all consistent with genuine, rapid organic adoption rather than artificially inflated metrics.

What Makes It Unique Most code-context tools for AI agents focus narrowly on source code; Graphify explicitly extends its knowledge graph to database schemas, infrastructure definitions, and non-code assets like PDFs, images, and video, aiming to represent “app code + database schema + infrastructure in one graph” as its own description puts it — a broader scope than typical code-search or embedding tools.

Self-Hosting

Licensing Model MIT licensed — fully open source with no license key.

Self-Hosting Restrictions None found; the skill/CLI runs locally or in MCP server mode against your own project files.

License Key Required No.