my_stock_decide_execute_skill

当且仅当用户明确要求“交易决策并执行”或提到“my_stock_decide_execute_skill”时触发。这是一个严格流程控制的编排型交易执行技能，强制按照“决策 -> 交易挂单 -> 日志记录”三步顺序执行，且每步都要求显式输出结果并由用户确认或自动串联。

Security Scan

The skill is internally coherent as an orchestration layer for making and placing real trades, but it enables high-risk actions (real-money order placement) while relying on external skills for auth and lacks strong, explicit safeguards against unintended autonomous execution.

Purpose & Capability

The name/description claim (orchestration: decision -> place order -> log) matches the SKILL.md instructions: it delegates decisioning to my_stock_decide_skill, order placement to my_stock_longbridge_skill, and logging to my_stock_log_skill. No unrelated binaries, env vars, or install steps are requested, which is proportionate for an orchestrator.

Instruction Scope

Instructions stay within the stated orchestration role and do not ask the agent to read system files or secrets. However, they require executing real trade placement via another skill and allow either explicit user confirmation or 'automatic chaining' between steps; the conditions for automatic chaining are vague and could enable orders without strong, programmatic confirmation safeguards.

Install Mechanism

Instruction-only skill with no install spec and no code files — lowest installation risk. Nothing is written to disk by this skill itself.

Credentials

This skill declares no credentials or env vars (appropriate for an orchestrator). It relies on other skills to perform authenticated actions; you should verify those downstream skills' credential requirements and ensure they are limited and stored/used securely. The orchestration role implies access to trade parameters which are sensitive.

Persistence & Privilege

always:false (good), but disable-model-invocation:false means the agent may autonomously invoke this skill. Combined with the ability to place real orders (via my_stock_longbridge_skill) and the SKILL.md allowance for 'automatic chaining', there is a real risk of unintended automated trades unless the agent/platform enforces the stated 'only trigger on explicit user request' and additional confirmation steps.

What to consider before installing

This skill is an orchestration controller that will call other skills to make trading decisions, place real orders, and write logs. Before installing, verify: (1) the downstream skills (my_stock_longbridge_skill, my_stock_decide_skill, my_stock_log_skill) actually exist and review their auth/credential needs; (2) there is a strict, auditable confirmation step before any real-money order is sent (prefer multi-factor or explicit user consent each time); (3) agent autonomous invocation is restricted if you do not want automatic execution — disable autonomous invocation at the platform level or require explicit user confirmations; (4) test end-to-end with a paper/training account to confirm behavior and logging; (5) ensure audit logs and error handling are implemented by the logging skill. If you cannot review or control the downstream skills' credential usage and confirmation logic, treat this skill as high risk and do not enable it for real trades.

Purpose (目的)

实现从“宏观/个股决策分析”到“长桥证券真实挂单”，再到“系统操作日志归档”的全自动化端到端闭环执行。

Core Rules & Triggers (触发条件与核心规则)

1. 触发条件：只有在用户输入中明确包含“交易决策并执行”字眼，或直接呼叫本技能名称“my_stock_decide_execute_skill”时，才允许启动本流程。严禁在常规闲聊或仅询问分析时触发实际交易。
2. 强制串行执行 (Sequential Execution)：流程必须严格按照以下三步进行。任何一步失败，必须立即终止并向老大报告，禁止跳步。
3. 分步显式输出 (Explicit Output per Step)：每一步执行完毕后，必须在返回给老大的消息中明确列出该步的输出结果。

Execution Flow (执行流程)

Step 1: 交易决策 (Decision Making)

• 动作：调用 my_stock_decide_skill，根据老大提供的持仓列表或特定标的进行分析。
• 强制输出：必须提取并向老大输出该技能的决策步骤及结论（包括：标的代码、操作方向买/卖、情绪、目标价格、目标仓位比例/股数，以及详尽的操作原因）。
• 等待确认/串联：确认决策结论已生成且包含有效操作建议。

Step 2: 交易操作 (Trade Execution)

• 动作：提取 Step 1 中的交易参数（代码、方向、价格、股数），严格按照这些参数调用 my_stock_longbridge_skill 进行实际的长桥证券挂单操作。
• 强制输出：必须向老大输出交易操作的挂单结果。这必须包括长桥返回的成功/失败状态，若是成功，需包含订单号（Order ID）或挂单回执；若是失败，需包含具体的错误原因（如余额不足、价格偏离过大等）。

Step 3: 日志记录 (Log Archiving)

• 动作：将 Step 1 的决策详情和 Step 2 的挂单结果（包含成功状态及订单号）整合，调用 my_stock_log_skill 写入系统的交易操作日志中。
• 强制输出：必须向老大输出“完成日志记录的条数”及简要的记录摘要（确认已成功入库）。

Security & Verification (安全与校验)

• 在执行 Step 2（实际交易）之前，智能体必须在内部逻辑上核实提取的参数完整且合法，绝不允许市价盲打（除非策略中明确要求市价单且参数正确）。
• 本技能作为编排控制器，不负责具体的策略计算、长桥 API 鉴权或数据库连接，只负责准确调度相关基础技能。