stock-scheme-tracker

Data & APIs

Save and track stock investment strategies with entry/exit conditions in a local Markdown file and check current market data to recommend next steps.

Install

openclaw skills install stock-scheme-tracker

Stock Strategy Tracker 📈

Save per-stock investment strategies extracted from analysis reports into a persistent tracking file, then periodically check if market conditions have triggered any entry, stop-loss, or take-profit conditions. Provide actionable next-step recommendations.


Quick Start

Install

Copy the skill directory to your agent's skills folder:

# OpenClaw
cp -r stock-scheme-tracker ~/.openclaw/skills/

# WorkBuddy
cp -r stock-scheme-tracker ~/.workbuddy/skills/

# Claude Code (project-level)
cp -r stock-scheme-tracker .claude/skills/

Configure

Set the environment variable to point to your tracking file:

# Add to your shell profile (~/.zshrc, ~/.bashrc, etc.)
export STOCK_SCHEME_PATH="$HOME/Documents/stock/scheme.md"

Or configure via your agent's config:

// OpenClaw: ~/.openclaw/openclaw.json
{
  "skills": {
    "entries": {
      "stock-scheme-tracker": {
        "env": {
          "STOCK_SCHEME_PATH": "~/Documents/stock/scheme.md"
        }
      }
    }
  }
}

If STOCK_SCHEME_PATH is not set, the skill defaults to ~/Documents/stock/scheme.md.

Verify

echo $STOCK_SCHEME_PATH
# Should output your configured path

Workflow Decision Tree

  1. Determine mode: Did the user just complete an analysis (→ save mode), or do they want to review existing strategies (→ check mode)?
  2. Save mode: Extract strategy elements from the analysis, format them, and append to the scheme file.
  3. Check mode: Read all strategies from the scheme file, fetch latest market data for each stock, evaluate every condition, and output a summary with next-step recommendations.

Save Mode

When to Use

  • Immediately after completing any stock investment analysis (e.g., DCF valuation, Buffett framework, technical analysis, or any investment report).
  • The user says "保存策略", "把策略记下来", "跟踪这个票", "save strategy", "track this stock", or any variant.

Extraction Rules

From the analysis text, extract these fields. If a field is not mentioned, mark it as "未明确" rather than fabricating.

FieldDescriptionExample
stock_codeStock code688111
stock_nameStock name金山办公
analysis_dateAnalysis date2026-04-22
recommendationCore recommendation建议观察,等待更好入场时机
entry_conditionsEntry conditions (price/valuation/events)PE≤45-50;股价180-200元
stop_lossStop-loss condition跌破前期低点或技术位
target_priceTarget price / take-profit range机构目标价400元
exit_signalsExit signals (assumption invalidation)月活环比下滑;市场份额跌破80%
position_sizeSuggested position size轻仓试探
time_horizonHolding period3年
key_assumptionsKey assumptionsAI商业化持续推进
open_questionsOpen questionsWPS AI付费转化率是多少
current_pricePrice at analysis~250元
current_pePE at analysis62.46
market_capMarket cap at analysis1147亿

Output Format

Append a new section to the scheme file (path from STOCK_SCHEME_PATH env var) using the exact format defined in references/scheme_format.md. Use Markdown table for entry conditions so they can be machine-parsed later.

Important

  • If the stock already exists in the scheme file, update the existing section rather than duplicating it. Preserve historical entries by appending a date-stamped update note.
  • After saving, briefly confirm to the user what was saved.

Check Mode

When to Use

  • User says "检查策略", "核对建仓条件", "看看哪些票可以买", "策略跟踪", "check strategy", or any variant.
  • Can be invoked standalone without any prior analysis.

Steps

  1. Read the scheme file (path from STOCK_SCHEME_PATH env var).
  2. For each tracked stock, fetch the latest market data (price, PE, key metrics). Use available financial data tools or web search.
  3. Evaluate each entry_conditions row: compare current value vs target value.
  4. Mark each condition as ✅ 已触发, ❌ 未触发, or ⚠️ 接近触发.
  5. Check if any exit_signals have occurred.
  6. Output a clean summary table and specific next-step recommendations per stock.

Output Format

## 策略核对报告 — 2026-04-23

### 金山办公 (688111)
| 条件类型 | 目标 | 当前 | 状态 |
|----------|------|------|------|
| PE建仓 | ≤45-50 | 62.46 | ❌ 未触发 |
| 股价建仓 | 180-200 | 250 | ❌ 未触发 |

**建议**: 继续观察,等待PE回落至50以下或股价跌破200再考虑建仓。

Core Tasks

Here are natural-language prompts that trigger this skill:

  • "我刚分析完金山办公,帮我保存策略"
  • "检查一下我跟踪的股票,哪些条件触发了"
  • "核对建仓条件"
  • "跟踪策略"
  • "把刚才的分析结论记录到策略跟踪文件"
  • "I just finished analyzing Apple, save the strategy"
  • "Check all my stock strategies"

Environment Variable Contract

VariablePurposeRequiredSet Via
STOCK_SCHEME_PATHPath to the strategy tracking Markdown fileNo (defaults to ~/Documents/stock/scheme.md)Shell profile or agent config

Security & Guardrails

  1. Local file only: Both scripts only read/write a local Markdown file specified by STOCK_SCHEME_PATH. No network calls are made by save_scheme.py or check_scheme.py.
  2. Market data source: During check mode, market data is fetched via the agent's available financial data tools or web search. This involves querying public market quotes (stock codes and metric names only) — no private strategy content or position details are transmitted.
  3. No trading execution: This skill provides analysis and recommendations only. It does NOT execute any trades or connect to brokerage accounts.
  4. No credential collection: This skill only requires STOCK_SCHEME_PATH (a file path). It never asks for API keys, tokens, passwords, or any other credentials.
  5. Data integrity: The save script uses atomic write patterns and preserves historical entries. Existing stock sections are updated in-place rather than overwritten.
  6. Path validation: The scripts validate that STOCK_SCHEME_PATH points to a .md file and will create parent directories if needed.

Troubleshooting

ProblemCauseFix
"No such file or directory"STOCK_SCHEME_PATH not set or points to invalid pathSet env var or rely on default ~/Documents/stock/scheme.md
Strategy not found during checkScheme file is empty or format is corruptedVerify the file uses the format in references/scheme_format.md
Duplicate entries for same stockSave mode ran twice without updatingThe script auto-detects duplicates; if using manual edit, follow the update convention
Chinese characters garbled in outputTerminal encoding mismatchEnsure your terminal uses UTF-8 (export LANG=en_US.UTF-8 or zh_CN.UTF-8)

Release Notes

  • 1.1.0 — Fixed metadata consistency (top-level requires/env), added credential collection disclaimer, clarified market data query scope in security section.
  • 1.0.0 — Initial release. Save and check modes with Markdown-based tracking file.

Links

Publisher