🌐 English · 简体中文 · 
AISEACT is an optional enhancement methodology, not a mandatory filter. You control when and how it's used.
| Benefit | Impact |
|---|---|
| 🔍 Higher Source Quality | P0 source usage: 5–15% → 85–100% |
| ⏱️ Time Efficiency | 75% less time on source verification |
| ✅ Better Accuracy | Fact-check error rate: −83% |
| 📋 Professional Research | Structured methodology for complex questions |
| 🔗 Traceability | Clear source labeling and citations |
| Feature | Description |
|---|---|
| 🔧 Invocation | Manual by default (explicit request required) |
| ⚙️ Autonomous Mode | Opt-in only; must be explicitly enabled |
| 🔄 Override | Always available; skip or customize anytime |
| 📖 Transparency | Source ratings are guidelines, not rules |
📖 Trust & Transparency Report → · Configuration Guide →
AISEACT provides a structured, optional methodology for improving AI search quality when you need rigorous, well-sourced answers.
✅ Recommended for:
- Academic or professional research
- Fact-checking and verification tasks
- Business or technical analysis where source quality matters
- Questions requiring traceable, authoritative sources
❌ Not needed for:
- Quick casual queries ("What's the weather?")
- Creative writing or brainstorming
- Subjective opinions or diverse perspectives
AI search often retrieves mixed-quality results:
- Content farms (百家号, scraper sites)
- Rewritten articles without original insight
- Unverifiable claims
AISEACT offers a solution: A systematic approach to prioritize primary sources and improve traceability — when you choose to use it.
Copy and paste this prompt to your AI agent (OpenCode, Claude Code, Kimi CLI, Cursor, etc.):
Install and configure aiseact skill by following the instructions here:
https://github.com/stephenlzc/aiseact
Or install manually:
# Clone the repository
git clone https://github.com/stephenlzc/aiseact.git
# Copy the entire directory to your AI agent's skills folder
# OpenCode
cp -r aiseact ~/.config/opencode/skills/
# Claude Code
cp -r aiseact ~/.claude/skills/
# Kimi CLI
cp -r aiseact ~/.kimi/skills/Strategy → Search → Analyze → Supplement → Validate → Answer
| Principle | Description |
|---|---|
| 📌 Primary Source Priority | Prefer official docs and original sources |
| 🔁 Iterative Search | Multiple rounds for complex questions |
| 🏷️ Source Transparency | Clear labeling of source types |
| 🎛️ User Control | You decide which sources to include |
[User] "Use AISEACT to research what makes Company X successful"
[AI with AISEACT]
Phase 0: Strategy Planning
└─ Decompose question, identify primary source locations
Phase 1: Initial Search
└─ Search for: Company X prospectus, annual reports
└─ Reference source quality guidelines
Phase 2: Analysis
└─ Identify missing information
Phase 3: Targeted Search
└─ site:cninfo.com.cn Company X prospectus
└─ site:company-x.com official data
Phase 4: Answer Generation
└─ Cite verifiable sources
└─ Label source types
└─ Note any uncertainties
Standard Search:
According to an online article, Company X IPO'd because... (Source: unspecified blog post)
With AISEACT:
According to Company X's prospectus [1], their IPO core advantages include:
- 2,000+ patents in battery technology
- Revenue of 19.997 billion RMB in 2017
Sources: [1] Company X Prospectus (Shanghai Exchange) — Primary Source
| Agent | Platform | Installation |
|---|---|---|
| 🦞 OpenClaw | Local/Cloud | openclaw skill install aiseact |
| 🐙 KimiClaw | Moonshot AI | Copy SKILL.md to skills folder |
| 🤖 Claude Code | Anthropic | Copy to .claude/skills/ |
| 📝 Cursor | Cursor AI | Copy to .cursor/skills/ |
# Clone the repository
git clone https://github.com/stephenlzc/aiseact.git
# Copy to your AI agent's skills folder
cp aiseact/SKILL.md ~/.claude/skills/Explicitly request AISEACT when you need it:
"请用AISEACT搜索..."
"Use AISEACT to research..."
"用AISEACT验证这个说法"
Enable automatic invocation in your configuration:
- Set
disable-model-invocation: falsefor this skill - Even enabled, you can override per-query:
"不用AISEACT，直接搜索..."
When AISEACT is active, you retain full control:
| Command | Effect |
|---|---|
"包含 [source]" |
Include a specific source regardless of ratings |
"排除 [source]" |
Exclude a specific source |
"显示所有来源" |
Don't filter; show all found sources |
"不用AISEACT" |
Skip the methodology for this query |
"用宽松模式" |
Reduce filtering strictness |
📖 See CONFIGURATION.md for more options.
Reference lists of commonly rated sources (see references/):
⚠️ Unreliable sources: Content farms, sites with low fact-check standards- ✅ Authoritative sources: Official documents, reputable media, academic sources
Important: These are reference materials based on third-party assessments. You can always request any source you need.
| Source Type | Description | Use Case |
|---|---|---|
| 📄 Primary | Original documents, official sources | Facts and data |
| 📑 Secondary | Analysis based on primary sources | Context and interpretation |
| 📰 Tertiary | Rewritten or aggregated content | Use with caution |
# Government info
policy_name site:gov.cn
# Company announcements (A-share)
company_name prospectus site:cninfo.com.cn
# Company announcements (Hong Kong)
company_name site:hkexnews.hk
# Technical docs
technology_name site:github.com filetype:pdfAISEACT integrates with multi-search-engine for powerful multi-engine search capabilities:
| Feature | Description |
|---|---|
| 17 Search Engines | 8 China-focused + 9 Global |
| No API Keys Required | Ready to use out of the box |
| Advanced Operators | site:, filetype:, time filters |
| Privacy Options | DuckDuckGo, Brave, Startpage |
| Knowledge Engine | WolframAlpha for calculations |
Engine Selection Guide:
| Query Type | Primary | Secondary |
|---|---|---|
| China policy | Baidu | Bing CN |
| Tech/Code | DuckDuckGo | |
| Privacy-sensitive | DuckDuckGo | Brave |
| Calculations | WolframAlpha | - |
| HK companies | Google HK | - |
| Global news | Bing |
Installation:
# Install multi-search-engine for OpenClaw
openclaw skill install https://clawhub.ai/gpyAngyoujun/multi-search-engine
# Or via LobeHub CLI
npx -y @lobehub/market-cli skills install mushroomfu-openclaw-skills-multi-search-engine📖 See references/multi-engine-guide.md for detailed usage.
aiseact/
├── README.md # This file
├── README.zh-CN.md # Chinese documentation
├── SKILL.md # Core methodology (for AI agents)
├── TRUST.md # Trust & transparency report
├── CONFIGURATION.md # Configuration guide
├── EVALUATION_REPORT.md # Performance evaluation report
├── SECURITY.md # Security documentation
└── references/
├── unreliable-sources.md # Source quality reference
├── authority-sources.md # Authoritative source recommendations
├── quick-reference.md # Bilingual quick reference
├── workflow.md # Detailed workflow guide
├── search-strategies.md # Search techniques
├── multi-engine-guide.md # Multi-search-engine integration
├── search-syntax.md # Unified search syntax reference
└── case-studies.md # Examples
aiseact/ ├── README.md # This file ├── README.zh-CN.md # Chinese documentation ├── SKILL.md # Core methodology (for AI agents) ├── TRUST.md # Trust & transparency report ├── CONFIGURATION.md # Configuration guide ├── EVALUATION_REPORT.md # Performance evaluation report ├── SECURITY.md # Security documentation └── references/ ├── unreliable-sources.md # Source quality reference ├── authority-sources.md # Authoritative source recommendations ├── quick-reference.md # Bilingual quick reference ├── workflow.md # Detailed workflow guide ├── search-strategies.md # Search techniques └── case-studies.md # Examples
---
## Trust & Transparency
### User Autonomy
- **Default:** Manual invocation only
- **Override:** Always available
- **Source Control:** You decide what to include
### Acknowledged Limitations
- **Western-centric bias** — Evaluation frameworks favor Western sources
- **Mainstream preference** — May undervalue independent/alternative sources
- **Temporal limitations** — Ratings are snapshots, not permanent
### Data Privacy
- No data storage
- No external servers beyond standard search APIs
- No credential requirements
📖 [Full Trust Report →](TRUST.md)
---
## Why AISEACT?
### Use Case: Research Quality
When you need:
- ✅ Verifiable claims with source URLs
- ✅ Primary sources over rewritten content
- ✅ Structured methodology for complex questions
AISEACT provides an optional framework — without forcing it on every query.
### Comparison
| Aspect | Standard Search | With AISEACT |
|--------|-----------------|--------------|
| Speed | Fast | Moderate (more thorough) |
| Source Quality | Mixed | Higher |
| User Control | Full | Full (optional) |
| Best For | Quick answers | Research, verification |
---
## Documentation
- [SKILL.md](SKILL.md) — Core methodology for AI agents
- [TRUST.md](TRUST.md) — Trust, transparency, and user control
- [CONFIGURATION.md](CONFIGURATION.md) — Customization options
- [EVALUATION_REPORT.md](EVALUATION_REPORT.md) — Performance metrics
- [references/workflow.md](references/workflow.md) — Detailed workflow
- [references/authority-sources.md](references/authority-sources.md) — Source recommendations
- [references/multi-engine-guide.md](references/multi-engine-guide.md) — Multi-search-engine integration
---
## 🙏 Acknowledgments
Special thanks to <img src="https://simpleicons.org/icons/tiktok.svg" width="16" height="16" alt="TikTok"/> **CHEN老师说AI** on Douyin for the inspiration and insights that helped shape this project.
Thanks to **[@gpyAngyoujun](https://clawhub.ai/gpyAngyoujun/multi-search-engine)** for the brilliant idea of aggregating 17 search engines into a unified skill — this concept sparked AISEACT's multi-engine research workflow. In the vibe coding era, **IDEAS ARE THE REAL ENGINE**.
---
## License
MIT © [stephenlzc](https://github.com/stephenlzc)
---
<div align="center">
**AISEACT** — Search Smarter, When You Need To 🔍
*Trust, verify, and always stay in control.*
[English](README.md) · [中文](README.zh-CN.md)
</div>