An automated bridge that transforms insight engine logs and learning feeds into a searchable, structured SQLite and LlamaIndex knowledge base.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install knowledge-bridge
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 knowledge-bridge 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 knowledge-bridge?
The knowledge-bridge skill functions as an automated data pipeline, designed to sync learning insights directly into a localized SQLite and LlamaIndex database. By extracting high-value information from online sources, such as video platforms, academic webinars, and research feeds, this skill acts as an intelligent layer that organizes raw digital intake into structured knowledge blocks. Users can leverage this mechanism to build a highly indexed, persistent personal brain without manual entry.
As part of the expanding ecosystem of Openclaw Skills, knowledge-bridge automatically categorizes incoming data by domains like AI/技术 (AI/Technology), 商业/变现 (Business/Monetization), and 教育/学习 (Education/Learning). It utilizes automated summarization to map video titles and URLs to distinct concepts, employing MD5-based deduplication to ensure your knowledge repository remains clean, accurate, and completely clutter-free.
knowledge-bridge Use Cases
- Bilibili & Online Learning Aggregation: Automatically compile watched video URLs, webinar titles, and key topics from educational platforms into your localized knowledge repository.
- Automated Research Ingestion: Pipe newly discovered papers, technical webinars (like VALSE_Webinar), and algorithmic insights straight into your personal indexing engine.
- Centralized Knowledge Querying: Search across your cumulative digital learning footprint using semantic search capabilities powered by LlamaIndex.
- Historical Knowledge Recovery: Run bulk sync operations to backfill past learning cycles and consolidate fragmented data from older log entries.
How knowledge-bridge Works
- Triggering: The workflow begins when a user invokes the skill with predefined phrases such as "知识桥", "knowledge bridge", or "ingest knowledge".
- Log Ingestion: The system reads recent learning entries, platform activity logs, and URLs from the primary script directory (
D:\\coze-local\\db\\knowledge_bridge.py). - Parsing & Enrichment: The pipeline processes metadata such as platform channel names, video URLs, and study timestamps, applying automated summaries and extraction.
- Deduplication: Every entry undergoes an MD5 hash verification to prevent duplicates in the knowledge base.
- Classification: The skill dynamically categorizes each insight into semantic domains (e.g., AI/Technology, Education, or Business).
- Storage & Indexing: Final entries are written to a localized SQLite database (
memory.db) and indexed utilizing LlamaIndex for advanced query retrieval.
knowledge-bridge Setup
To get started with knowledge-bridge and incorporate it into your collection of Openclaw Skills, follow these configuration steps:
- Ensure your local Python environment has the necessary dependencies installed:
pip install llamaindex sqlite3
Configure the file paths in your environment settings or directly within your local database config module to point to your target log directories:
- Primary python script location:
D:\\coze-local\\db\\knowledge_bridge.py - SQLite Database file path:
D:\\coze-local\\logs\\eye_log\\memory.db
- Primary python script location:
Test the ingestion command locally:
python D:\\coze-local\\db\\knowledge_bridge.py --action ingest
knowledge-bridge Data Schema & Taxonomy
The knowledge-bridge database structure maps high-fidelity metadata from incoming learning cycles into structured SQLite fields.
Core Database Schema (memory.db)
| Table/Entity | Field | Type | Description |
|---|---|---|---|
| Knowledge Entries | id |
INTEGER (PK) | Unique primary key for the record |
title |
TEXT | Title of the source video or article (e.g., VALSE_Webinar) | |
url |
TEXT | Direct URL (e.g., Bilibili video path) | |
category |
TEXT | Domain category (AI/技术, 商业/变现, 教育/学习) | |
summary |
TEXT | Synthesized auto-summary or keywords extracted from title | |
hash_id |
TEXT | MD5 hash used for ingestion deduplication | |
timestamp |
DATETIME | Exact timestamp of the learning cycle ingest |
Directory Layout
- Execution File:
D:\\coze-local\\db\\knowledge_bridge.py— Coordinates core ingestion and search routing logic. - Storage Engine:
D:\\coze-local\\logs\\eye_log\\memory.db— Stores raw SQLite table schemas alongside embedded vector matrices created by LlamaIndex.
knowledge-bridge Advanced Features
- MD5 Deduplication Filter: Automatically detects duplicate entries across multiple learning cycles to prevent database bloat and preserve index sanity.
- Semantic Search Engine: Combines standard SQL keyword queries with semantic vector matching via LlamaIndex for intelligent retrieval of indexed topics.
- Multi-Cycle Backfilling: Supports historical bulk-loading (
run(\"backfill\")) allowing users to index their entire historical catalog of Bilibili records in a single run. - Multi-Domain Taxonomy: Out-of-the-box support for domain-specific categorization targeting high-value learning vectors like machine learning algorithms (e.g., Diffusion Schrodinger Bridge) and technical webinars.
SKILL.md
Loading
Related Openclaw Skills
Quick Draft
A productivity skill that transforms brief conversational prompts and key points into polished, structured email and message drafts.
JSON Transformer
An auto-crystallized utility for querying, merging, and filtering JSON datasets with self-evolving execution capabilities.
Habit Check-in
An automated habit tracking and check-in skill for AI agents that monitors daily streaks, historical records, and total active days.
Agently Mail
An elegant CLI-driven skill to manage Tencent QQMail email workflows, covering search, reads, sends, and attachments via terminal commands.
MCP Tool Maker
An automated development skill that scans Python scripts and auto-generates Model Context Protocol (MCP) compliant server tools.
Skill Distillation
A method and toolchain for extracting, formulating, and distilling reusable AI agent skills from LLMs, human expertise, and structural methodologies.
Featured*
