feat: AI Bounty Description Enhancer with multi-LLM analysis and maintainer approval workflow

[zengbinli]

AI Bounty Description Enhancer

Implements an AI agent that analyzes vague bounty descriptions and generates improved versions with clearer requirements, acceptance criteria, and examples.

Features

• Multi-LLM analysis: Queries Claude, GPT, and Gemini in parallel
• Frequency-based aggregation: Consensus items from multiple LLMs rank highest
• Confidence scoring: Each provider self-assesses improvement quality
• Maintainer approval workflow: ENHANCED → APPROVED/REJECTED pipeline
• CLI tool: Standalone command-line interface
• FastAPI routes: Ready for backend integration
• Comprehensive tests: 13 unit tests covering aggregation, approval, parsing, and models

Components (scripts/bounty_enhancer/)

• models.py — Type-safe dataclasses for all inputs/outputs
• providers.py — Multi-LLM integration (Claude/GPT/Gemini/DeepSeek) via OpenAI-compatible API
• enhancer.py — Core engine with aggregation and approval workflow
• cli.py — Command-line interface
• api_routes.py — FastAPI endpoints for backend integration

Aggregation Strategy

1. Title: Highest confidence suggestion wins
2. Description: Most detailed version (with sanity check)
3. Requirements/Criteria: Union with dedup, ranked by agreement frequency
4. Examples: All unique examples merged

Usage

# CLI
python -m scripts.bounty_enhancer enhance "Fix login bug" "Users can't log in"
python -m scripts.bounty_enhancer enhance-file bounty.yaml
python -m scripts.bounty_enhancer approve ID -r maintainer

Testing

python -m pytest tests/bounty_enhancer/ -v  # 13 tests

Closes #848

Wallet: 0x3032c11E48E9479fB65F0Cf8bEceCd570aB6b800