StockBuddy

Data and caching behavior: - Raw daily K-line data, watchlist data, and portfolio data are stored in ~/.stockbuddy/stockbuddy.db (SQLite) - Positions are linked through watchlist_id - Analysis results are cached separately in SQLite with a default TTL of 10 minutes - Cache cleanup runs automatically and total cached analysis rows are capped - If the user explicitly asks to "refresh data" or "reanalyze", add --no-cache - To clear analysis cache: --clear-cache

3. Interpret and present the result - The script returns JSON analysis data - For default single-stock requests, use the default query template in references/output_templates.md - The default response must include: stock basics, data-driven action recommendation (with score and confidence), important events, event-adjusted second-pass suggestion, and practical order ideas - Default order style = balanced. Only switch when the user explicitly asks for a conservative or aggressive version - Only produce the full report when explicitly requested with phrases like "full report", "detailed analysis", or "complete analysis" - The top natural-language summary is mandatory in both short and long versions: 2-4 sentences covering regime, main recommendation, confidence, support/risk points, and whether the stock is actionable today - Only expand into a more open-ended explanation when the user asks for detail such as "explain why", "show the reasoning", "how about short-term", or "what stop-loss/stop-profit should I use" - Final output must be normal Markdown, not wrapped in code fences; prefer short paragraphs, bullet points, and card-style formatting over wide tables unless the user explicitly wants a detailed report

Scenario 2: Batch Portfolio Analysis

Trigger examples: "analyze my portfolio", "look at my holdings", "how are my positions doing"

Default output should still be decision-first: for each position, give the action label, score/confidence, important events, event-adjusted second suggestion, and a compact practical order version. Do not expand every holding into a full long report unless the user explicitly wants a detailed version.

Steps:

1. Check portfolio data

   python3 {{SKILL_DIR}}/scripts/portfolio_manager.py list
   

2. If the portfolio is empty → guide the user to add positions first (see Scenario 3)

3. Run batch analysis

   python3 {{SKILL_DIR}}/scripts/portfolio_manager.py analyze
   

4. Interpret and present the result - Format the result using the "Portfolio Batch Analysis Report" section in references/output_templates.md - Output normal Markdown, not code fences - It may use standard Markdown tables mixed with lists when helpful, but keep it readable on chat surfaces - Include stock-level recommendations and portfolio-level P&L summary - Prefer real-time computed fields in the output: latest price, market value, unrealized P&L, position weight, and executable action constraints such as whole-lot vs odd-lot behavior - Do not write latest price, position weight, or unrealized P&L back into durable storage; the database should only hold stable user-confirmed facts and trading rules

Scenario 3: Position Management

Trigger examples: "add a Tencent position", "I bought 100 shares of BYD", "remove Alibaba from my holdings"

| Action | Command | |------|------| | Add position | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py add --price --shares [--date ] [--note ] [--account ] | | List positions | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py list | | Update position | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py update [--price ] [--shares ] [--note ] [--account ] | | Remove position | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py remove | | List accounts | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py account-list | | Create/update account | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py account-upsert [--market ] [--currency ] [--cash ] [--available-cash ] [--note ] | | Set trading rule | python3 {{SKILL_DIR}}/scripts/portfolio_manager.py rule-set [--lot-size ] [--tick-size ] [--odd-lot] |

When adding a position, ensure the stock exists in the watchlist and is linked through positions.watchlist_id -> watchlist.id. If the user does not provide a date, default to the current date. If the user provides natural-language trade info such as "I bought 100 shares of Tencent last week at 350", extract price, share count, date, and account info where possible, then execute the appropriate command.

After the first successful position record, proactively guide the user to fill the missing durable facts needed for execution-aware advice. Do not stop at only code / shares / cost if important constraints are still unknown.

Ask for or help the user confirm these fields, in this priority order: 1. Account context — which account this belongs to, and its market / currency 2. Cash — total cash or available cash in that account 3. Lot rule — lot size for the stock 4. Odd-lot support — whether the broker supports odd-lot selling / buying 5. Trade date — if omitted and relevant for later review 6. Notes — optional thesis, time horizon, or special constraints

If some fields are already known, only ask for the missing ones. Keep the follow-up compact: confirm what was captured, state what is still missing, and ask for the missing durable facts in one short message.

If the user gives only partial follow-up info, update what is available and continue asking only for the remaining missing fields. Once enough execution constraints are known, stop prompting and proceed normally.

Scenario 4: Account and Allocation Management

Trigger examples: "my HK account has 3000 HKD cash", "track available cash", "record this under my US account", "how concentrated is my portfolio"

Rules: