Introduction

The core design philosophy of DeepSeek Harness (DSH) is a plugin-based architecture. For developers who are accustomed to using Obsidian for knowledge management (PKM), integrating their note vault into DSH’s agent workflows is a common requirement. The dsh-obsidian-second-brain plugin is ported from the obsidian-second-brain project and adapted to DSH’s plugin model. It aims to turn an Obsidian vault into a second brain that can maintain itself, organize automatically, and perform self-validation.

Plugin Overview

This plugin is maintained by GongYuanCaiJi and is licensed under the MIT License. Its core goal is to automate the manual process of organizing notes: information from new sources updates existing pages rather than merely appending to them, contradictions between notes are automatically reconciled, and outdated facts are flagged to prevent the knowledge base from becoming increasingly cluttered.

Core Features

  1. Self-Rewriting Knowledge Base
    Newly input information attempts to update existing note entries instead of simply adding content at the bottom. The plugin automatically reconciles contradictory information in the vault.

  2. 46 Commands Covering the Complete Workflow
    The plugin provides 46 commands, covering every aspect of a second brain, including capture, daily notes, projects, people, tasks, kanban boards, decisions, retrospectives, retrieval, and health checks.

  3. Hybrid Semantic Retrieval
    It uses title-weighted retrieval and supports self-evaluation of retrieval quality to ensure the accuracy of the information.

  4. Research Toolbox
    It provides 7 research commands (including X, web pages, YouTube, podcasts, and NotebookLM). Research results are automatically saved to the vault.

  5. Scheduled Maintenance
    The plugin supports automatic vault cleanup at session end and during background agent runs.

  6. AI-First Writing Rules
    Every generated note includes structured frontmatter, wikilinks, source information, and confidence levels, designed specifically for future AI retrieval rather than merely human readability.

  7. Typed Graph
    Relationships between notes such as supersedes (supersedes), depends_on (depends on), and contradicts (contradicts) are defined. These relationships can be verified through scripts.

Installation

To install the plugin in a DSH environment, use the following command:

dsh plugin --profile <你的 profile> add github:GongYuanCaiJi/dsh-obsidian-second-brain

If pnpm intercepts the build step, add this package to allowBuilds in the profile’s pnpm-workspace.yaml file.

Usage

After installation, you can use the obsidian-second-brain skill in a DSH session. The skill guides your operations based on the state of your vault:

  1. Skill Guidance: At the start of a session, the obsidian-second-brain skill appears in the list of available skills.
  2. Read Rules: The plugin preferentially reads the _CLAUDE.md file in the vault root directory (if it exists) as the operating rules for the current operation.
  3. Save Dialogues: Use the /obsidian-save command to organize the current dialogue into interlinked AI-first notes (including entries for people, projects, decisions, tasks, and more).
  4. Retrieve Notes: Use the /obsidian-find command to search for notes in the vault.

Note: Research commands (such as /research, /x-read, etc.) depend on the upstream Python research tooling (pyproject.toml). uv will be prompted for installation on first use.

Notes and Dependencies

  • Language Environment: The upstream content is in English. Command trigger words include multiple languages (including Chinese); per the porting rules, they are not translated and remain unchanged.
  • Python Dependencies: The research features depend on Python and the uv package manager.
  • MCP Bridging: The MCP bridge files (integrations/obsidian-mcp-server/) are kept in their original upstream location and enabled as needed.
  • File Fidelity: The plugin preserves all 321 upstream files verbatim and pins commit 4d5b673. All command definitions and scripts remain consistent with the upstream.

Summary

dsh-obsidian-second-brain is an Obsidian plugin adaptation designed for AI-first knowledge management. Through structured writing rules and automated maintenance mechanisms, it helps DSH users interact efficiently with Obsidian in agent sessions, enabling continuous accumulation and updating of knowledge.

  • Directory Page: https://www.skillhub.cn/plugins/GongYuanCaiJi/dsh-obsidian-second-brain
  • GitHub: https://github.com/GongYuanCaiJi/dsh-obsidian-second-brain