logical-chain-search

Universal multi-round search using logical-chain reasoning for any AI coding agent.

What It Does

Replaces stateless single-shot search (Result = Search(Query)) with a stateful Markov decision process: Query_{n+1} = Logic(Result_n, Context, Goal).

Each search iteration builds on the previous round's results, connected by one of five explicit logical operators:

Operator	Symbol	Direction
Drill-down	🔍	Macro → Micro
Trace-back	🔙	Effect → Cause
Pivot	↔️	Blocked → Alternative
Cross-Verify	⚖️	External → Internal validation
Fill-gap	🧩	Known unknowns → Targeted retrieval

Use Cases

• Code debugging: Trace root cause through multi-layer systems
• Architecture research: Evaluate technology choices with evidence chains
• Technology evaluation: Compare alternatives with structured reasoning
• Root cause analysis: Systematic investigation of production issues
• Competitive analysis: Deep-dive into solution landscapes

Installation

Google Antigravity (Gemini)

git clone https://github.com/linliu0210/logical-chain-search.git

Then copy or symlink:

# Skill
cp -R logical-chain-search/ /path/to/your/project/.agent/skills/logical-chain-search/

# Workflow (optional — for /deep-search slash command)
cp logical-chain-search/workflow/deep-search.md /path/to/your/project/.agent/workflows/deep-search.md

Claude Code

cp -R logical-chain-search/ ~/.claude/skills/logical-chain-search/

Cursor

cp -R logical-chain-search/ .cursor/rules/logical-chain-search/

GitHub Copilot

cp -R logical-chain-search/ .github/skills/logical-chain-search/

Usage

Invoke with /deep-search followed by your investigation topic:

/deep-search 排查这个 P99 延迟飙升的原因
/deep-search investigate why the build fails on CI but passes locally
/deep-search 对比 Kafka vs Pulsar 在低延迟场景下的选型
/deep-search trace down the memory leak in the worker pool

File Structure

logical-chain-search/
├── SKILL.md                          # Core engine (cognitive loop, operators, dispatch)
├── references/
│   ├── logical-operators.md          # Detailed operator reference with examples
│   └── research-board.md             # Working memory schema & quality criteria
├── workflow/
│   └── deep-search.md                # Antigravity /deep-search workflow
└── README.md                         # This file

License

MIT