Agent测试
- 作者仓库星标 1,568
- 作者更新于 实时读取
- 作者仓库 claude-trading-skills
- 领域
- 数据
- 兼容 Agent
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- 信任分
- 88 / 100 · 社区维护
- 作者 / 版本 / 许可
- @tradermonty · 未声明 license
- Token 消耗评级
- 低消耗
- 接入复杂程度
- 需简单配置
- 是否需要外部 API Key
- 不需要
- 兼容的系统
- 未声明(默认跨平台)
- 底层运行要求
- Python >=3.9
- 文件与系统权限
-
- 只读
- 允许写入 / 修改
- Shell 执行
- 网络行为
- 仅限本地
- 安装命令数
- 26 条
档案由构建时根据 SKILL.md 与安装命令自动衍生,可能与作者实际意图存在差异。
需要注意: 未限定 allowed-tools,默认拥有全部工具权限。
---
name: skill-integration-tester
description: Validate multi-skill workflows defined in CLAUDE.md by checking skill existence, inter-skill dat…
category: 数据
runtime: Python
---
# skill-integration-tester 输出预览
## PART A: 任务判断
- 适用问题:表格、CSV、数据集、指标或分析流程。
- 输入要求:目标材料、限制条件、期望输出和验收方式。
- 证据边界:围绕“Overview / When to Use / Prerequisites”读取原文规则,不把推断写成作者承诺。
## PART B: 执行结果
- **01** 任务判断:确认你的需求是否属于表格、CSV、数据集、指标或分析流程,并标出输入、限制和预期结果。
- **02** 执行计划:优先按“Overview / When to Use / Prerequisites”拆成步骤,说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果:给出可复制的命令、文件改动、检查清单或内容草稿,并说明如何继续迭代。
- **04** 风险边界:结合 读取文件、写入/修改文件、执行终端命令、主要在本地完成、通常不需要额外 API Key 给出执行前确认项。
## Running Rules
- 读取文件、写入/修改文件、执行终端命令;主要在本地完成;通常不需要额外 API Key。
- 先小样例验证,再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。 原文没有稳定的斜杠命令要求。安装验证后通常全局生效,直接在对话里点名这个 Skill 并描述任务即可。
告诉 Agent 目标文件或材料、期望结果、不可改范围、是否允许联网或执行命令。本 Skill 的权限画像是:读取文件、写入/修改文件、执行终端命令。
先用一个小任务确认它会围绕“Overview / When to Use / Prerequisites”工作;涉及文件或命令时,先看 diff、日志、预览或测试结果。
检查最终产物是否包含明确结果、必要证据和下一步动作;如果输出泛泛而谈,就补充输入、边界和验收标准后重跑。
---
name: skill-integration-tester
description: Validate multi-skill workflows defined in CLAUDE.md by checking skill existence, inter-skill dat…
category: 数据
source: tradermonty/claude-trading-skills
---
# skill-integration-tester
## 什么时候使用
- 用于组织测试、定位失败并形成修复闭环 适合处理表格、CSV、指标、数据集、分析和可视化报告,核心价值是把输入、判断、执行、验证和交付边界固定下来,避免 Agent 泛泛回答。 把任务拆成可执行、可检查、可继续迭代的步骤;通常不需要额外…
- 面向表格、CSV、数据集、指标或分析流程,优先处理能明确输入、步骤和验收标准的工作。
## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。
## 执行规则
- 围绕「Overview / When to Use / Prerequisites」组织步骤,不把推断写成作者事实。
- 读取文件、写入/修改文件、执行终端命令;主要在本地完成;通常不需要额外 API Key。
- 先跑小样例,确认结果可检查后再扩大任务范围。
## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown,不编造命令、平台或依赖。 作者原文负责流程事实;仓库文件负责来源和命令;流狐只补充适用场景、限制和质量判断。
skill "skill-integration-tester" {
输入层 -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
上下文层 -> Overview / When to Use / Prerequisites
规则层 -> SKILL.md 触发条件 / 执行顺序 / 输出格式
运行层 -> Python | 读取文件、写入/修改文件、执行终端命令 | 主要在本地完成
安全层 -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
输出层 -> 可复制结果 + 检查清单 + 下一步迭代
} Skill Integration Tester
Overview
Validate multi-skill workflows defined in CLAUDE.md (Daily Market Monitoring, Weekly Strategy Review, Earnings Momentum Trading, etc.) by executing each step in sequence. Check inter-skill data contracts for JSON schema compatibility between output of step N and input of step N+1, verify file naming conventions, and report broken handoffs. Supports dry-run mode with synthetic fixtures.
When to Use
- After adding or modifying a multi-skill workflow in CLAUDE.md
- After changing a skill's output format (JSON schema, file naming)
- Before releasing new skills to verify pipeline compatibility
- When debugging broken handoffs between consecutive workflow steps
- As a CI pre-check for pull requests touching skill scripts
Prerequisites
- Python 3.9+
- No API keys required
- No third-party Python packages required (uses only standard library)
Workflow
Step 1: Run Integration Validation
Execute the validation script against the project's CLAUDE.md:
python3 skills/skill-integration-tester/scripts/validate_workflows.py \
--output-dir reports/
This parses all **Workflow Name:** blocks from the Multi-Skill Workflows
section, resolves each step's display name to a skill directory, and validates
existence, contracts, and naming.
Step 2: Validate a Specific Workflow
Target a single workflow by name substring:
python3 skills/skill-integration-tester/scripts/validate_workflows.py \
--workflow "Earnings Momentum" \
--output-dir reports/
Step 3: Dry-Run with Synthetic Fixtures
Create synthetic fixture JSON files for each skill's expected output and validate contract compatibility without real data:
python3 skills/skill-integration-tester/scripts/validate_workflows.py \
--dry-run \
--output-dir reports/
Fixture files are written to reports/fixtures/ with _fixture flag set.
Step 4: Review Results
Open the generated Markdown report for a human-readable summary, or parse the JSON report for programmatic consumption. Each workflow shows:
- Step-by-step skill existence checks
- Handoff contract validation (PASS / FAIL / N/A)
- File naming convention violations
- Overall workflow status (valid / broken / warning)
Step 5: Fix Broken Handoffs
For each FAIL handoff, verify that:
- The producer skill's output contains all required fields
- The consumer skill's input parameter accepts the producer's output format
- File naming patterns are consistent between producer output and consumer input
Output Format
JSON Report
{
"schema_version": "1.0",
"generated_at": "2026-03-01T12:00:00+00:00",
"dry_run": false,
"summary": {
"total_workflows": 8,
"valid": 6,
"broken": 1,
"warnings": 1
},
"workflows": [
{
"workflow": "Daily Market Monitoring",
"step_count": 4,
"status": "valid",
"steps": [...],
"handoffs": [...],
"naming_violations": []
}
]
}
Markdown Report
Structured report with per-workflow sections showing step validation, handoff status, and naming violations.
Reports are saved to reports/ with filenames
integration_test_YYYY-MM-DD_HHMMSS.{json,md}.
Resources
scripts/validate_workflows.py-- Main validation scriptreferences/workflow_contracts.md-- Contract definitions and handoff patterns
Key Principles
- No API keys required -- all validation is local and offline
- Non-destructive -- reads SKILL.md and CLAUDE.md only, never modifies skills
- Deterministic -- same inputs always produce same validation results
先判断是否适合
作者设计意图
作者的方法与取舍
边界和复核