Content
# Tool List
**AI-powered Content Creation Engine** - Transforming input into output
[](https://www.python.org/)
[](LICENSE)
[](https://modelcontextprotocol.io/)
## One-sentence Introduction
**Spend 5 minutes a day, let Knowledge Hunter transform input into output**
> You've followed 50 bloggers, scrolled through Twitter/Weibo...
> But these contents are forgotten after reading, never becoming your own content.
Knowledge Hunter helps you: **Monitor → Transcribe → Extract key sentences → AI score → Generate draft**
## Core Features
### 🤖 AI Capability Scheduling
Supports multiple AI service providers, automatic switching:
| Provider | Model | Applicable Scenario |
|--------|------|----------|
| **OpenAI** | GPT-4o / GPT-4o-mini | General generation |
| **DeepSeek** | deepseek-chat | High cost-effectiveness |
| **Claude** | Claude-3.5-Sonnet | High-quality long text |
| **Moonshot** | moonshot-v1-128k | Ultra-long context |
| **Local Whisper** | base/small | Free transcription |
### ⚡ Complete Workflow
```
📥 Input Layer 🔧 AI Scheduling Layer 📤 Output Layer
───────── ────────── ─────────
• Podcast RSS • Transcription (multi-model) • Key sentence library
• YouTube • Scoring (0-10 points) • Topic list
• Bilibili • Extraction (viewpoints/key sentences) • Draft creation
• Twitter • Generation (multi-format) • Daily briefing
• RSS Subscription
```
### 🔌 MCP Plugin Support
Can be directly called by AI Agent (Claude Code / OpenClaw / others):
```json
{
"mcpServers": {
"knowledge-hunter": {
"command": "python",
"args": ["/path/to/scripts/mcp_server.py"]
}
}
}
```
**23 Available Tools**: Search key sentences, intelligent recommendation, create tasks, generate drafts, AI scoring, scheduling management...
## Quick Start
### 1. Installation
```bash
git clone https://github.com/chubbyxiaopangdun/knowledge-hunter.git
cd knowledge-hunter
pip install -r requirements.txt
```
### 2. Configure AI Keys (select at least one)
```bash
export OPENAI_API_KEY=sk-xxx
# or
export DEEPSEEK_API_KEY=sk-xxx
# or
export ANTHROPIC_API_KEY=sk-ant-xxx
```
### 3. Run
```bash
# Interactive configuration wizard
python scripts/setup/wizard.py
# Monitor content
python scripts/monitor.py
# Creative calendar
python scripts/creative_calendar.py --today
# One-click generate draft
python scripts/one_click_generator.py --topic "AI entrepreneurship" --format thread
# Start task scheduling (background automatic execution)
python scripts/scheduler/cli.py init --force
python scripts/scheduler/cli.py start
```
## Functional Modules
| Module | File | Description |
|------|------|------|
| **Creative Calendar** | `creative_calendar.py` | Daily task management |
| **One-click Generation** | `one_click_generator.py` | Thread/ Xiaohongshu/ WeChat Official Account |
| **Quick Search** | `quick_search.py` | Key sentence search and copy |
| **AI Scoring** | `ai_scorer.py` | 5-dimension scoring 0-10 |
| **MCP Server** | `mcp_server.py` | 23 Agent tools |
| **Intelligent Deduplication** | `deduplicator.py` | URL/title/content triple deduplication |
| **Interactive Wizard** | `setup/wizard.py` | Configuration guidance |
## MCP Tool List
### Key Sentence Management
- `search_quotes` - Search key sentences
- `add_quote` - Add key sentence
- `score_quote` - AI scoring
- `toggle_favorite` - Favorite/unfavorite
### Task Management
- `list_tasks` - List tasks
- `create_task` - Create task
- `start_task` / `complete_task` - Status update
- `add_draft` - Add draft
### AI Capabilities
- `generate_draft` - Generate draft
- `extract_from_text` - Extract key sentences
- `summarize_text` - Summarize text
### Scheduler Management
- `get_scheduler_status` - Get scheduler status
- `list_scheduled_tasks` - List scheduled tasks
- `pause_scheduled_task` / `resume_scheduled_task` - Pause/resume task
## Task Scheduling
### Automatic Task Execution
```bash
# Initialize scheduler (register default tasks)
python scripts/scheduler/cli.py init --force
# View scheduling status
python scripts/scheduler/cli.py status
# Start scheduler (background running)
python scripts/scheduler/cli.py start
# View task list
python scripts/scheduler/cli.py list
```
### Predefined Tasks
| Task | Trigger | Description |
|------|------|------|
| Daily full monitoring | Daily 8:00 | Execute full content monitoring |
| RSS incremental monitoring | Hourly | Check RSS updates |
| Daily briefing | Daily 9:00 | Generate creative briefing |
| Key sentence scoring | Daily 22:00 | AI scoring for unscored key sentences |
| Obsidian synchronization | Daily 23:00 | Synchronize to knowledge base |
## Directory Structure
```
knowledge-hunter/
├── scripts/
│ ├── ai/ # AI scheduling layer
│ │ ├── ai_manager.py # Core scheduler
│ │ └── adapters/ # Model adapters
│ ├── config/ # Unified configuration management
│ │ ├── __init__.py
│ │ └── models.py # Configuration data model
│ ├── scheduler/ # Task scheduling engine
│ │ ├── models.py # Task data model
│ │ ├── store.py # SQLite storage
│ │ ├── scheduler.py # APScheduler core
│ │ ├── integration.py # Predefined tasks
│ │ └── cli.py # Command-line tool
│ ├── platform/ # Platform monitoring
│ │ └── rss_monitor.py
│ ├── setup/
│ │ └── wizard.py # Configuration wizard
│ ├── creative_calendar.py # Creative calendar
│ ├── one_click_generator.py # One-click generation
│ ├── quick_search.py # Quick search
│ ├── ai_scorer.py # AI scoring
│ ├── mcp_server.py # MCP Server (23 tools)
│ └── deduplicator.py # Content deduplication
├── config/ # Configuration files
├── data/ # Knowledge base data
├── docs/
│ └── MCP Installation Guide.md # Agent access guide
└── requirements.txt
```
## Version History
| Version | Date | Update |
|------|------|------|
| **v2.2** | 2026-05 | Intelligent recommendation + natural language search + scheduling notification (23 MCP tools) |
| **v2.1** | 2026-05 | Centralized configuration + task scheduling engine |
| **v2.0** | 2026-05 | AI scheduling layer + MCP plugin support |
## Frequently Asked Questions
**Q: Do I need to pay?**
A: Basic features are free. Transcription can use local Whisper (free), AI capabilities require your own API Key.
**Q: Which AI models are supported?**
A: OpenAI, DeepSeek, Claude, Moonshot, local Whisper, extensible.
**Q: How to let Agent use?**
A: Configure MCP Server, Agent can directly call all tools.
**Q: Where is the data stored?**
A: Local JSON files, no privacy risk.
## License
MIT License
<p align="center">
<a href="https://github.com/chubbyxiaopangdun/knowledge-hunter">GitHub</a> ·
<a href="https://github.com/chubbyxiaopangdun/knowledge-hunter/issues">Feedback</a>
</p>
