Run any Skill in Manus with one click

Get Started

innovate-poc

Stars0

Forks0

UpdatedFebruary 28, 2026 at 03:32

創新 Phase 3：POC 原型實作。產出可執行的 proof-of-concept 程式碼。Use when 'build POC', 'prototype', '做原型', 'proof of concept'.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

danny0926

danny0926/NLP-data-for-trading

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software DevelopersComputer and Mathematical Occupations·SOC 15-1252

SKILL.md

readonly

name	innovate-poc
description	創新 Phase 3：POC 原型實作。產出可執行的 proof-of-concept 程式碼。Use when 'build POC', 'prototype', '做原型', 'proof of concept'.
agent	general-purpose
allowed-tools	Read, Glob, Grep, Bash, Edit, Write, Task

/innovate-poc — POC 原型實作

讀 docs/innovation_state.json 取得掃描和映射結果，自主建立 POC。

POC 原則

最小可行 — 只實作核心邏輯
可測量 — 必須產出可比較的數字（CAR, hit rate, alpha vs SPY）
可重現 — 記錄所有參數和資料版本
Anti look-ahead bias — 使用 filing_date, benchmark=SPY

Step 1: 建立 POC 目錄

poc/
└── YYYY-MM-DD-<topic>/
    ├── README.md          # 說明與結論
    ├── poc_<topic>.py     # 主程式
    ├── requirements.txt   # 額外依賴（如有）
    └── results/           # 輸出結果

Step 2: 寫 POC 程式碼

"""
POC: [Topic]
Date: [YYYY-MM-DD]
Author: PAM Innovation Engine

Hypothesis: [一句話]
Expected: [預期結果]
Anti look-ahead bias: [如何確保]
"""
import sqlite3
import pandas as pd
from src.config import DB_PATH

# ===== CONFIG =====
TOPIC = "[topic]"
DB = DB_PATH
BENCHMARK = "SPY"
EVENT_DATE_COL = "filing_date"  # 必須用 filing_date

# ===== DATA =====
conn = sqlite3.connect(DB)
trades = pd.read_sql("SELECT * FROM congress_trades WHERE ticker IS NOT NULL", conn)

# ===== IMPLEMENTATION =====
# ... minimal core logic ...

# ===== EVALUATION =====
# Must output: CAR_5d, CAR_20d, hit_rate, sample_size, p_value
# Compare with baseline: RB-001 Buy CAR_5d = +0.77%

Step 3: 執行測試

# 確保不破壞現有測試
pytest tests/ -v

# 執行 POC
cd poc/YYYY-MM-DD-<topic>/
python poc_<topic>.py

Step 4: 記錄結果

寫 poc/YYYY-MM-DD-<topic>/README.md：

# POC: [Topic]

## Hypothesis
[一句話]

## Setup
- Data: congress_trades from data/data.db
- Period: [start] to [end]
- Sample: [N] trades
- Benchmark: SPY

## Results
| Metric | Value | vs Baseline |
|--------|-------|-------------|
| CAR_5d | X.XX% | [+/-] vs RB-001 0.77% |
| CAR_20d | X.XX% | [+/-] vs RB-001 0.79% |
| Hit Rate | X% | [+/-] vs 55% baseline |
| p-value | X.XXX | [sig/not sig] |

## Conclusion
[2-3 句]

## Bias Compliance
- [x] Event date = filing_date
- [x] Benchmark = SPY
- [x] No future information used
- [x] Sample size >= 30

Step 5: 輸出

更新 docs/innovation_state.json 的 poc phase 為 done。

## POC Report: [Topic]

- Directory: `poc/YYYY-MM-DD-<topic>/`
- Key metric: [CAR/hit_rate/etc.] = [value]
- Baseline comparison: [+/- delta vs RB-001]

### Next: `/innovate-verdict`

More from this repository

same repository

supervise

danny0926/NLP-data-for-trading

自主 CEO 模式。代替使用者監管整個 Agent Team，自主決策研發方向、派工、監控、處理異常。支援 2+ 小時 headless 運行，跨 context window 自動接力。Use when 'supervise', '監管', '自主運行', 'autopilot', '代替我管', 'run team', '幫我盯', '工作幾小時'.

2026-03-070

innovate-map

danny0926/NLP-data-for-trading

創新 Phase 2：跨域映射 + 數據源評估。分析技術遷移可行性與新數據源 alpha 潛力。Use when 'cross-domain mapping', '跨域映射', 'data source evaluation', '數據源評估'.

2026-02-280

innovate-scan

danny0926/NLP-data-for-trading

創新 Phase 1：方向偵測 + 前沿掃描。搜論文、GitHub、社群找最新方法。Use when 'frontier scan', '搜論文', 'explore new tech', '前沿掃描'.

2026-02-280

innovate

danny0926/NLP-data-for-trading

創新 Pipeline 指揮官。前沿探索 → 跨域映射 → 數據源評估 → POC 原型 → 可行性裁定。可一鍵跑完或逐步執行。Use when 'innovate', '創新', '新方法', '探索', 'what if we try', '新數據源', '新想法'.

2026-02-280

innovate-verdict

danny0926/NLP-data-for-trading

創新 Phase 4：可行性裁定、評分、記錄到 innovation_log、銜接後續。Use when 'innovation verdict', '裁定', 'feasibility score', '可行性評估'.

2026-02-280

research-analyze

danny0926/NLP-data-for-trading

研究 Phase 4：回測結果分析、A/B 比較、偏差快篩。Use when 'analyze results', '分析結果', 'compare backtest'.

2026-02-280

/innovate-poc — POC 原型實作

讀 docs/innovation_state.json 取得掃描和映射結果，自主建立 POC。

POC 原則

最小可行 — 只實作核心邏輯

可測量 — 必須產出可比較的數字（CAR, hit rate, alpha vs SPY）

可重現 — 記錄所有參數和資料版本

Anti look-ahead bias — 使用 filing_date, benchmark=SPY

Step 1: 建立 POC 目錄

poc/ └── YYYY-MM-DD-<topic>/ ├── README.md # 說明與結論 ├── poc_<topic>.py # 主程式 ├── requirements.txt # 額外依賴（如有） └── results/ # 輸出結果

Step 2: 寫 POC 程式碼

""" POC: [Topic] Date: [YYYY-MM-DD] Author: PAM Innovation Engine Hypothesis: [一句話] Expected: [預期結果] Anti look-ahead bias: [如何確保] """ import sqlite3 import pandas as pd from src.config import DB_PATH # ===== CONFIG ===== TOPIC = "[topic]" DB = DB_PATH BENCHMARK = "SPY" EVENT_DATE_COL = "filing_date" # 必須用 filing_date # ===== DATA ===== conn = sqlite3.connect(DB) trades = pd.read_sql("SELECT * FROM congress_trades WHERE ticker IS NOT NULL", conn) # ===== IMPLEMENTATION ===== # ... minimal core logic ... # ===== EVALUATION ===== # Must output: CAR_5d, CAR_20d, hit_rate, sample_size, p_value # Compare with baseline: RB-001 Buy CAR_5d = +0.77%

Step 3: 執行測試

# 確保不破壞現有測試 pytest tests/ -v # 執行 POC cd poc/YYYY-MM-DD-<topic>/ python poc_<topic>.py

Step 4: 記錄結果

寫 poc/YYYY-MM-DD-<topic>/README.md：

# POC: [Topic] ## Hypothesis [一句話] ## Setup - Data: congress_trades from data/data.db - Period: [start] to [end] - Sample: [N] trades - Benchmark: SPY ## Results | Metric | Value | vs Baseline | |--------|-------|-------------| | CAR_5d | X.XX% | [+/-] vs RB-001 0.77% | | CAR_20d | X.XX% | [+/-] vs RB-001 0.79% | | Hit Rate | X% | [+/-] vs 55% baseline | | p-value | X.XXX | [sig/not sig] | ## Conclusion [2-3 句] ## Bias Compliance - [x] Event date = filing_date - [x] Benchmark = SPY - [x] No future information used - [x] Sample size >= 30

Step 5: 輸出

更新 docs/innovation_state.json 的 poc phase 為 done。

## POC Report: [Topic] - Directory: `poc/YYYY-MM-DD-<topic>/` - Key metric: [CAR/hit_rate/etc.] = [value] - Baseline comparison: [+/- delta vs RB-001] ### Next: `/innovate-verdict`