datit309/supergraph

A workflow system that makes a coding agent behave like a disciplined engineer — evidence-gated scan → analyze → plan → TDD → execute → fix → verify → review, backed by a local codebase-memory graph and optional Serena LSP.

Supergraph is a workflow system for AI coding agents — not a prompt collection — that enforces planning, TDD, verification, review and architecture-aware decision making. Every non-trivial change follows scan -> analyze -> plan -> TDD -> execute -> fix -> verify -> review, and 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. Its graph layer is powered by Codebase Memory MCP, which indexes the repository locally under a stable project identity and supplies blast radius, callers/callees, architecture clusters, dependency cycles, test gaps, complexity hotspots and changed symbols; Serena is optional for LSP references. Hooks provide session reminders, plan guards, failure hints and asynchronous graph freshness. The README also documents Windows hook behaviour: hooks/run-hook.cmd resolves Git Bash dynamically and prints supergraph: Git Bash not found — hooks skipped (exit 0, non-blocking) when it is unavailable, leaving skills and MCP usable.

Workflow & Automation ★ 22
View on GitHub ↗

Install

dsh plugin --profile web add github:datit309/supergraph

Copied verbatim from the README's Option 5 (DeepSeek Harness), Option A (plugin bundle, recommended). A local-checkout form (dsh plugin --profile web add /path/to/supergraph) and a Web GUI install (Settings -> Plugins -> Install Bundle, enter github:datit309/supergraph) are also documented. The update command is dsh plugin --profile web update supergraph.

Compatibility

Standard Cordis Plugin & Bundle integration for DeepSeek Harness. Requires Python 3.10+ and the codebase-memory-mcp package (>=0.10.8,<0.11.0); Serena MCP is optional and adds LSP-level references and diagnostics. The DSH bundle auto-registers 38 skills via ctx.skills.registerProvider(), configures the codebase-memory and serena MCP servers through @deepseek-ai/dsh-mcp-client, and injects workflow rules from AGENTS.md into ctx.systemPrompt.

Details

Alternatives

GooDAnDReaDY/dsh-cron · GooDAnDReaDY/dsh-goal · gosomea/dsh-task-supervisor

More plugins in Workflow & Automation

Browse more in Workflow & Automation

Guides for Workflow & Automation plugins