将本地 flowix-cli 注册为 MCP 服务,让 agent 可以搜索、读取、创建和编辑 Flowix 备忘与思维导图产物。
安装
# GitHub 源码(首次需按提示配置 allowBuilds 构建授权后重试)
dsh plugin --profile web add github:text2future/flowix#path:/dsh-flowix-memory
装任何插件都等于在你的机器上跑第三方代码,权限和你本人一样大——能读你的文件、用你的凭据、访问网络,工具审批管不到它。GitHub 来源的插件还会在安装时执行构建脚本——pnpm 默认拦截,所以安装可能停在 ERR_PNPM_GIT_DEP_PREPARE_NOT_ALLOWED 或 ERR_PNPM_IGNORED_BUILDS;dsh 会打印出需要添加的确切键名,把它加进该 profile 的 pnpm-workspace.yaml 的 allowBuilds 下,重跑一次即可装上。放行构建本身就是一次信任判断:请只安装可信来源,并尽量锁定 commit(github:owner/repo#sha)。
README
该插件的 README 只有英文版本。
A DeepSeek Harness bundle
that registers the local flowix-cli MCP server with any Harness
instance. Once installed, the agent gets the
mcp__flowix__memo tool to search, read, create, and edit
Flowix Markdown memos, and to create declared plugin artifacts such as mind
maps.
The bundle includes a small launcher. It transparently proxies to flowix-cli
when installed; otherwise it keeps the tool registered and returns an
actionable installation error to the agent. The launcher is a cordis.patch.yml
that inserts one
@deepseek-ai/dsh-mcp-client
row, which ships inside every Harness distribution. Nothing here connects to a
remote Flowix service — the CLI is spawned locally over stdio and reads the
local notebook data directory.
Prerequisites
- A
flowixcommand onPATH(or setFLOWIX_CLI_PATHto the absoluteflowix-clibinary path) is needed to use the memo operations. If it is missing, the tool explains how to install Flowix fromhttps://flowix-memo.com/latest.json. The CLI is built from the flowix-main repository (app/flowix-cli, binaryflowix-cli). - Access to the Flowix notebook data the CLI should manage. Defaults to the
user config directory (
~/.flowix); override withFLOWIX_HOME/FLOWIX_DATAwhen the data lives elsewhere.
Install
dsh plugin manages one profile at a time and requires --profile <name>.
Flowix uses the flowix profile. In other DSH clients, flowix is simply a
custom profile name and can be created by the same command. The bundle remains independent from
Flowix's host/control bridge and ships in the
flowix-main repository and is not yet
published to npm, so install it from the checkout:
# from the flowix-main checkout root
dsh plugin --profile flowix add ./dsh-flowix-memory
Once published, the npm package installs the same way:
dsh plugin --profile flowix add dsh-flowix-memory
The row activates for that profile. Other DSH clients can start it with
dsh --profile flowix; this profile also ships dsh-appserver so the
Flowix App Server JSON-RPC surface is available out of the box.
Verify
Start Harness with the profile you installed into (dsh --profile flowix),
then check that the tool is registered:
mcp__flowix__memo
Prefer structured arguments:
{"action":"search","query":"product plan","notebook":"work","tag":"project/flowix","limit":20}
The legacy { "command": "search ...", "stdin": "..." } form remains
accepted for compatibility but is no longer the recommended contract.
Configuration
| Key | Default | Meaning |
|---|---|---|
FLOWIX_CLI_PATH |
flowix (PATH lookup) |
Absolute path to the flowix-cli executable |
If the CLI is unavailable, calling the tool returns an installation message
with https://flowix-memo.com/latest.json. Install Flowix, then restart the
profile so the launcher reconnects to flowix-cli mcp.
Uninstall
dsh plugin --profile flowix remove dsh-flowix-memory
Notes
- The MCP stdio bridge strips ambient credential-like variables and all
DSH_*variables before spawning the child; other ambient variables (includingHOME/PATH) are inherited. - Flowix Desktop installs this same bundle into its
flowixprofile and setsFLOWIX_DSH_MCP_CLI. Other DSH clients can install it into any profile and useFLOWIX_CLI_PATHor theflowixexecutable on PATH. There is one canonical bundle patch and no embedded duplicate row.
链接
同类插件
vectorize-io/hindsight#coding-agents★ 48490
Hindsight:会学习的 Agent 长期记忆系统,自动召回/保存、知识页、深度反思与按仓库隔离的记忆银行。
volcengine/OpenViking#examples/dsh-memory-plugin★ 39688
面向 DeepSeek Harness 的 OpenViking 记忆与上下文插件:pre-step 自动召回与画像注入、会话捕获、`viking://` URI 防护,以及对接 OpenViking 服务端的 recall/write 记忆工具。
agentscope-ai/ReMe#dsh★ 3578
将 DeepSeek Harness 接入 ReMe 本地优先、自进化的个人知识库:自动把已完成的主 Agent 对话沉淀为用户掌控的 Markdown 记忆,通过 reme_search 结合 BM25、可选向量检索和 wikilink 展开搜索对话与资料,并按日整理长期记忆。
zilliztech/memsearch#MemSearch★ 2736
供 DSH 与其他编程 Agent 共享的 Markdown 记忆,支持自动捕获、步骤前上下文注入、搜索召回,以及通过审阅面板实现 memory-to-skill 自进化。
vshulcz/deja-vu#extensions/dsh★ 1176
读取本机上其他三十三个编程智能体已经写下的会话文件——Claude Code、Codex、Cursor、VS Code Copilot Chat、opencode、OpenClaw、Hermes、Kimi、Cline、Zed 等,包括安装之前的历史:提供 deja_recall、deja_session、deja_blame、deja_fix、deja_how、deja_remember 六个工具与 /deja 命令,并可选地把召回结果加入运行时上下文。本地 BM25 索引,不用大模型,不用向量嵌入,不联网(dsh plugin --profile web add dsh-deja)。
adoresever/graph-memory★ 644
DeepSeek Harness 的可追溯、可检索跨会话记忆:把对话知识沉淀为带类型的图节点(任务/技能/事件)与关系边。
社区评论
评论公开保存在 GitHub Discussions。加载评论会连接 GitHub 和 Giscus;发表内容需要 GitHub 账号。