Tooluck logoTooluck
SKILL.md

investor-panel

Agent skill fromwbh604/UZI-Skill

66 位投资大佬评审团。给定一只股票的 dimensions.json 和 raw_data.json,让 66 位投资者各自按自己的方法论打分并输出 Pydantic Signal(signal/confidence/score/verdict/comment)。覆盖经典价值派、成长投资派、宏观对冲派、技术趋势派、中国价投派、A股游资派、量化系统派、科技领袖派、AI 卡位猎手 9 大流派。当用户请求"评审团/65 大佬怎么看/某某会买吗/做一次大佬投票"时使用。

How to install investor-panel

install investor-panel
# install the UZI-Skill pack with the skills CLI
npx skills add wbh604/UZI-Skill

# or add only this skill to Claude Code: copy the folder

skills/investor-panel/

# into

~/.claude/skills/investor-panel/ (all projects)

.claude/skills/investor-panel/ (one project)

skills/investor-panel/SKILL.md

MITFull file
1# Investor Panel · 50 贤评审团
2
3## 调用上下文
4
5读取以下输入:
6- `.cache/{ticker}/dimensions.json` — 19 维评分
7- `.cache/{ticker}/raw_data.json` — 原始数据
8- `scripts/lib/investor_db.py` — 65 人元数据
9- `scripts/lib/seat_db.py` — 22 位游资射程规则
10
11输出:
12- `.cache/{ticker}/panel.json` — 50 个 Signal + 投票统计
13
14## 严格输出格式(Pydantic Signal,抄自 ai-hedge-fund)
15
16每个投资者必须返回**严格 JSON**:
17
18```json
19{
20 "investor_id": "buffett",
21 "name": "巴菲特",
22 "group": "A",
23 "avatar": "avatars/buffett.svg",
24 "signal": "bullish | neutral | bearish",
25 "confidence": 87,
26 "score": 82,
27 "verdict": "强烈买入 | 买入 | 关注 | 观望 | 等待 | 回避 | 不达标 | 不适合",
28 "reasoning": "1-3 句具体逻辑",
29 "comment": "用该投资者语言风格的金句 1-2 句",
30 "pass": ["..."],
31 "fail": ["..."],
32 "ideal_price": 16.20,
33 "period": "3-5 年"
34}
35```
36
37**Confidence 校准规则**:
38- 85-100:核心方法论硬指标全部命中或全部不命中
39- 60-84:多数命中
40- 30-59:部分命中、需要等待信号
41- 0-29:方法论不适用此股 / 信息不足
42
43## 执行步骤
44
45### Step 1: 加载元数据
46```python
47from lib.investor_db import INVESTORS, by_group
48from lib.seat_db import SEATS, is_in_range
49```
50
51### Step 2: 对每位投资者
521. 取出 `fields` 白名单
532. 从 dimensions.json 提取相关字段
543. 读取该投资者所在 group 的 reference 文件(按需)
554. 用该投资者的方法论 + 语言样本生成 Signal(Claude 自己生成)
565. 校验 JSON 合法性
57
58### Step 3: 游资射程预过滤(F 组特殊)
59对 22 位游资,先用 `is_in_range(nickname, ticker_features)` 判断是否在射程内:
60- 在射程 → 正常评分
61- 不在射程 → `signal: "neutral"`, `verdict: "不适合"`, `confidence: 90`, `comment: "{nick}的射程是{style},这只票不在风格内。"`
62
63### Step 4: 汇总投票
64```python
65{
66 "panel_consensus": (bullish_count / 50) * 100,
67 "vote_distribution": Counter(verdict for i in investors),
68…

Excerpt shown under the repository's MIT license.

More skills in UZI-Skill

Keep exploring

Similar skill repositories

thedotmack /

claude-mem
116k

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

CCCXGHGM+135 skills

nexu-io /

open-design
107k

The 1st Collaborative AI Design Agent Workspace. Local-first and open source. Create motion, prototypes, websites, apps, dashboards, presentations, images, and videos with your team and AI agents. Share work, exchange feedback, and iterate together in one workspace.

40 skills

tt-a1i /

archify
95k

Turn any idea, plan, or codebase into a beautiful interactive diagram. An agent skill for Claude Code, Codex, and more.

CCCXOC1 skill

Newsletter

Join the Community

Subscribe to our newsletter for the latest news and updates