---
title: "graphflow"
description: "图谱上下文压缩与任务规划插件（10 个 MCP 工具）。能力：压缩上下文、DAG 规划、增量建图、技能飞轮、诊断与产物导入导出。DeepSeek Harness：dsh plugin --profile web add @roarpeng/graphflow；工具名为 mcp__graphflow__graphflow_*。先调 graphflow_context（传 rootDir），复杂任务再 plan，改完代码后 index。"
canonical: https://agentpluginsdirectory.com/plugins/graphflow
last-updated: 2026-09-25
---

# graphflow
图谱上下文压缩与任务规划插件（10 个 MCP 工具）。能力：压缩上下文、DAG 规划、增量建图、技能飞轮、诊断与产物导入导出。DeepSeek Harness：dsh plugin --profile web add @roarpeng/graphflow；工具名为 mcp__graphflow__graphflow_*。先调 graphflow_context（传 rootDir），复杂任务再 plan，改完代码后 index。
- Slug: graphflow
- Publisher: Roarpeng
- Repository: https://github.com/Roarpeng/GraphFlow
- Manifest: plugin.json
- Version: 1.26.0
- License: Apache-2.0
- Category (editorial): agent-tooling
- Skills: 1 (graphflow)
- MCP servers: 1 (graphflow)
- Stars: 13
- Repository created: 2026-05-27
- Repository last pushed: 2026-09-23
- Publisher type: User
- Listing: https://agentpluginsdirectory.com/plugins/graphflow
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What graphflow does, in the publisher's words

> The memory & context harness for coding agents. Local-first code knowledge graph · bounded context compression (~98% token savings) · cross-session learning flywheel.

The community is converging on an "agent harness" vocabulary: memory + hooks + skills are the harness primitives that turn a stateless model into a reliable long-running agent. GraphFlow implements all three for coding agents and ships them through a portable MCP surface (Cursor, Claude Code, 15+ agents):

Pure TypeScript/Node. CLI + MCP + VS Code extension. Fully offline, no API key required.

- Static injection pays the same token cost every session regardless of the task, and grows until it is truncated or ignored.
- Plain RAG retrieves text but never accumulates experience �?the thousandth task pays the same cost as the first.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/Roarpeng/GraphFlow/HEAD/README.md

## Skills

- graphflow: 图谱上下文压缩、任务规划与知识图谱编排（10 个 MCP 工具）。任何读代码、改代码、排错、中文问题之前必须先调 graphflow_context。DeepSeek Harness 下工具名为 mcp__graphflow__graphflow_*。

Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.

## MCP servers

- graphflow: transport: stdio; command: npx -y --package=@roarpeng/graphflow graphflow-mcp; env: GRAPHFLOW_MCP_STDIO, GRAPHFLOW_LOG_JSON

Read from the plugin's own mcp.json. Environment variable names only, never values.