Supergraph

Supergraph for AI Coding Agents

[Tiếng Việt](./README-VI.md)

**Turn AI coding agents from code generators into engineering workflow systems.**

SuperGraph doesn't make your coding agent smarter. It makes your coding agent behave like a disciplined engineer.

SuperGraph enforces planning, TDD, verification, review, and architecture-aware decision making through mandatory workflows, graph intelligence, and LSP-powered code analysis.

What the plugin does

Supergraph is a workflow system for AI coding agents, not just a collection of prompts. Every non-trivial change follows:

`scan → analyze → plan → TDD → execute → fix → verify → review`

The graph layer is powered by [Codebase Memory MCP](https://github.com/DeusData/codebase-memory-mcp) (`>= 0.9.0`). It indexes the repository locally under a stable project identity, then provides evidence for blast radius, callers/callees, architecture clusters, dependency cycles, test gaps, complexity hotspots, and changed symbols. [Serena](https://github.com/oraios/serena) is optional and adds LSP-level references and diagnostics.

The workflow is evidence-gated: no production change without a verified RED test, no plan execution without an approved plan, and no completion claim without fresh verification and independent review. Hooks provide session reminders, plan guards, failure hints, and asynchronous graph freshness through Codebase Memory auto-watch.

Windows hook behavior

On Windows, `hooks/run-hook.cmd` resolves Git Bash dynamically: `CLAUDE_CODE_GIT_BASH_PATH`, system Git, user-level Git, then `where git.exe`. This supports Git for Windows installed by winget without administrator rights. If Git Bash is unavailable, hooks print `supergraph: Git Bash not found — hooks skipped` and exit successfully because hooks are non-blocking; skills and MCP remain usable.

[](./plugins/supergraph/CHANGELOG.md) [![License](https://img.shields.io/badge/license-M