MCP HubMCP Hub
スキル一覧に戻る

Decision Framework

derekcrosslu
更新日 Today
60 閲覧
0
GitHubで表示
開発general

について

意思決定フレームワークCLIは、バックテスト、最適化、検証の結果を評価することで、戦略開発のための自律的な意思決定を可能にします。事前に定義された基準に基づいて、プロジェクトを適切な次のフェーズに自動的に振り分けます。開発者は主要なプロジェクトフェーズ完了後に本CLIを使用し、継続、中止、または上位判断を要する決定を行うべきです。

クイックインストール

Claude Code

推奨
プラグインコマンド推奨
/plugin add https://github.com/derekcrosslu/CLAUDE_CODE_EXPLORE
Git クローン代替
git clone https://github.com/derekcrosslu/CLAUDE_CODE_EXPLORE.git ~/.claude/skills/Decision Framework

このコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします

ドキュメント

Decision Framework CLI

Evaluate results and route to next phase: venv/bin/python SCRIPTS/decision_cli.py (use as decision)

When to Load This Skill

  • After backtest completes (Phase 3 decision)
  • After optimization completes (Phase 4 decision)
  • After validation completes (Phase 5 decision)
  • Need to route to next phase

CLI Commands (Progressive Disclosure)

Evaluate Backtest (Phase 3)

# Evaluate backtest results
venv/bin/python SCRIPTS/decision_cli.py evaluate-backtest \
  --results PROJECT_LOGS/backtest_result.json \
  --state iteration_state.json

# JSON output
venv/bin/python SCRIPTS/decision_cli.py evaluate-backtest --results backtest.json --json

Decisions: PROCEED_TO_OPTIMIZATION | PROCEED_TO_VALIDATION | ABANDON_HYPOTHESIS | ESCALATE_TO_HUMAN

Evaluate Optimization (Phase 4)

# Evaluate optimization results
venv/bin/python SCRIPTS/decision_cli.py evaluate-optimization \
  --results PROJECT_LOGS/optimization_result.json \
  --state iteration_state.json

Decisions: PROCEED_TO_VALIDATION | USE_BASELINE_PARAMS | ESCALATE_TO_HUMAN | PROCEED_WITH_ROBUST_PARAMS

Evaluate Validation (Phase 5)

# Evaluate validation results
venv/bin/python SCRIPTS/decision_cli.py evaluate-validation \
  --results PROJECT_LOGS/validation_result.json \
  --state iteration_state.json

Decisions: DEPLOY_STRATEGY | PROCEED_WITH_CAUTION | ABANDON_HYPOTHESIS | ESCALATE_TO_HUMAN

Route to Next Phase

# Determine next action based on decision
venv/bin/python SCRIPTS/decision_cli.py route \
  --phase backtest \
  --decision PROCEED_TO_OPTIMIZATION \
  --iteration 1

Workflow

  1. Run Phase: Execute backtest/optimization/validation
  2. Evaluate: decision evaluate-<phase> --results results.json
  3. Route: decision route --phase <phase> --decision <DECISION>
  4. Execute Next: Proceed to next phase based on routing

Decision Thresholds

Loaded from iteration_state.json (single source of truth):

  • performance_criteria.minimum_viable - Sharpe 0.5, DD 0.35, Trades 20
  • performance_criteria.optimization_worthy - Sharpe 0.7, DD 0.30, Trades 30
  • performance_criteria.production_ready - Sharpe 1.0, DD 0.20, Trades 50
  • overfitting_signals.too_perfect_sharpe - Sharpe > 3.0
  • overfitting_signals.too_few_trades - Trades < 10

Do not hardcode thresholds. Always read from iteration_state.json.

Progressive Disclosure Pattern

Load only what you need:

  • Phase 3: Use evaluate-backtest (only backtest logic loaded)
  • Phase 4: Use evaluate-optimization (only optimization logic loaded)
  • Phase 5: Use evaluate-validation (only validation logic loaded)

Before (old approach):

  • Load 500-line decision-framework skill
  • Load 300-line backtesting-analysis skill
  • Total: 800 lines for any decision

After (CLI approach):

  • Run decision evaluate-backtest (instant, 100-line skill)
  • Progressive disclosure: 87.5% context reduction

Authoritative Documentation

When confused about decision logic or thresholds:

  • Read: PREVIOUS_WORK/PROJECT_DOCUMENTATION/autonomous_decision_framework.md
  • Contains: Complete decision tree, all thresholds, routing logic

Do not guess thresholds. Use authoritative docs as source of truth.

CLI Help

Use --help for command details:

venv/bin/python SCRIPTS/decision_cli.py --help
venv/bin/python SCRIPTS/decision_cli.py evaluate-backtest --help
venv/bin/python SCRIPTS/decision_cli.py route --help

IMPORTANT: Do not read decision_cli.py source code unless strictly needed for debugging. Use --help for usage.


Context Savings: 100 lines (vs 800 lines loading multiple skills) = 87.5% reduction

Progressive Disclosure: Load only the evaluation logic you need (backtest vs optimization vs validation)

Trifecta: CLI works for humans, teams, AND agents

Beyond MCP Pattern: Use --help, not source code. Load only what you need.

GitHub リポジトリ

derekcrosslu/CLAUDE_CODE_EXPLORE
パス: .claude/skills/decision-framework

関連スキル

algorithmic-art

メタ

This Claude Skill creates original algorithmic art using p5.js with seeded randomness and interactive parameters. It generates .md files for algorithmic philosophies, plus .html and .js files for interactive generative art implementations. Use it when developers need to create flow fields, particle systems, or other computational art while avoiding copyright issues.

スキルを見る

subagent-driven-development

開発

This skill executes implementation plans by dispatching a fresh subagent for each independent task, with code review between tasks. It enables fast iteration while maintaining quality gates through this review process. Use it when working on mostly independent tasks within the same session to ensure continuous progress with built-in quality checks.

スキルを見る

executing-plans

デザイン

Use the executing-plans skill when you have a complete implementation plan to execute in controlled batches with review checkpoints. It loads and critically reviews the plan, then executes tasks in small batches (default 3 tasks) while reporting progress between each batch for architect review. This ensures systematic implementation with built-in quality control checkpoints.

スキルを見る

cost-optimization

その他

This Claude Skill helps developers optimize cloud costs through resource rightsizing, tagging strategies, and spending analysis. It provides a framework for reducing cloud expenses and implementing cost governance across AWS, Azure, and GCP. Use it when you need to analyze infrastructure costs, right-size resources, or meet budget constraints.

スキルを見る