A privacy-first, zero-dependency local search engine that enables AI agents to retrieve historical context from markdown memories.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install reminiscence
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install reminiscence using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Zhuiyi: Local Semantic Memory Search?
Zhuiyi is a sophisticated local semantic search skill designed to empower AI agents with long-term memory retrieval. It operates entirely on your local machine without the need for external API keys or heavy vector databases, ensuring your data remains private and secure. By integrating this into Openclaw Skills, your agent can recall past conversations, project notes, and daily logs with high precision using a blend of traditional information retrieval and modern signal fusion.
The skill is optimized for performance and portability, written to run on the Python standard library. It transforms flat markdown files like MEMORY.md and daily logs into a searchable index, allowing the agent to answer questions such as "What did we discuss last week?" or "Find my notes on project X" instantly.
Zhuiyi: Local Semantic Memory Search Use Cases
- Searching through historical logs and daily diaries for specific events or decisions.
- Recalling context from long-term memory during complex coding or research tasks.
- Automatically retrieving relevant information when a user mentions "I remember..." or "Search for...".
- Providing the AI agent with a structured way to access personal knowledge bases stored in markdown format.
How Zhuiyi: Local Semantic Memory Search Works
- The system checks for the existence of a local index file located at ~/.openclaw/memory_bm25_index.json.
- If the index is missing or an update is requested, it parses the MEMORY.md file and the memory/ directory to build a fresh index.
- Upon receiving a query, it performs a coarse ranking using the BM25 algorithm to identify candidate documents.
- It then applies a multi-dimensional reranking process, factoring in IDF-Coverage, N-gram proximity, and exact phrase matching.
- Results are deduplicated and returned with metadata, including file paths, line numbers, and content abstracts for the agent to process.
Zhuiyi: Local Semantic Memory Search Setup
To build or refresh your memory index, execute the following command:
python3 ~/.openclaw/workspace/skills/Zhuiyi/scripts/search.py --build
To manually search your memories from the command line, use:
python3 ~/.openclaw/workspace/skills/Zhuiyi/scripts/search.py "your search query"
Zhuiyi: Local Semantic Memory Search Data Schema & Taxonomy
The skill organizes data into a JSON-based index for rapid retrieval. The schema follows this structure:
| Attribute | Description |
|---|---|
| Index File | ~/.openclaw/memory_bm25_index.json |
| Data Sources | MEMORY.md, memory/*.md |
| Tokenization | CJK 2-6 gram (Python 3.13 compatible) |
| Record Metadata | file path, start_line, end_line, chunk_idx, raw text |
| Scoring Signals | BM25norm (38%), IDF-Coverage (22%), N-gram Proximity (20%), ExactPhrase (12%), Soft Dice (10%) |
Zhuiyi: Local Semantic Memory Search Advanced Features
- Multi-signal fusion ranking that combines word frequency with structural proximity for higher accuracy.
- Zero-dependency architecture, relying solely on Python standard libraries for maximum compatibility with Openclaw Skills.
- Robust CJK support, making it highly effective for searching East Asian languages via character-based N-grams.
- Automatic deduplication logic to ensure that the agent receives the most relevant and concise context fragments.
SKILL.md
Loading
Related Openclaw Skills
Guangzhou City Guide
A dedicated AI agent skill for exploring Guangzhou's rich history, iconic landmarks, and world-renowned Cantonese cuisine.
Cherry Blossom Trip
A professional AI agent skill for planning cherry blossom trips with real-time booking data for Japan, Wuhan, and global sakura destinations.
Luxury Hotel & Travel Search
A powerful CLI-driven skill for AI agents to discover, compare, and book elite 5-star hotels and luxury travel accommodations.
WeChat SVG Effects
A specialized toolkit for building interactive, JavaScript-free SVG and CSS animations tailored for WeChat Official Account articles.
Warren Buffett Investment OS
A practical investment framework and cognitive operating system modeled after Warren Buffett’s 60-year track record of value investing.
Pilot Webhook Bridge
A powerful utility to forward Pilot Protocol events to external webhooks like Slack, Discord, and PagerDuty for real-time monitoring.
Featured*
