← 返回 Skills 市场
Run Test Plan
作者
Kevin Anderson
· GitHub ↗
· v1.0.1
· MIT-0
98
总下载
0
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install run-test-plan
功能描述
Execute YAML test plan, stop on first failure, output rich debug prompt
安全使用建议
This skill is a legitimate test-runner but can execute any shell command embedded in the YAML plan and will import environment variables from the running environment. Before using it: (1) review the test plan file yourself for any dangerous commands or network calls; (2) run it in an isolated environment or CI runner that has no sensitive credentials or secrets mounted; (3) remove or sanitize any environment variables you don't want exposed (DATABASE_URL, AWS_*, SSH keys, etc.); (4) prefer running only plans from trusted sources; and (5) if you must run untrusted plans, restrict network access and filesystem permissions for the runner process.
能力评估
Purpose & Capability
Name and description match the instructions: the SKILL.md describes parsing a YAML test plan, running setup, health checks, and sequential tests (shell, curl, agent-browser). The agent-browser dependency and use of curl, nohup, and file-based evidence/pid logs are expected for a test runner.
Instruction Scope
The instructions tell the agent to execute arbitrary shell commands taken directly from the YAML plan and to export environment values resolved from the current environment. There is no explicit restriction or sanitization of those commands or network endpoints. The skill also writes logs/pids to .beagle/ and evidence to docs/testing/evidence/. Executing arbitrary commands from an untrusted plan can run anything on the host.
Install Mechanism
Instruction-only skill with no install spec and no code files — nothing will be downloaded or written by an installer. This minimizes install-time risk.
Credentials
The skill explicitly resolves ${VAR} references from the current environment when exporting setup.env values but declares no required environment variables. That mismatch means it will read whatever env vars are present (including secrets like DATABASE_URL, AWS_*, etc.) without documenting them. Tests run by the plan could also access files or network resources and thereby exfiltrate secrets.
Persistence & Privilege
always is false and disable-model-invocation is true (it cannot autonomously call the model). The skill only writes service logs/pids and evidence under repository-style paths (.beagle/, docs/testing/evidence/) and does not attempt to change other skills or global agent settings.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install run-test-plan - 安装完成后，直接呼叫该 Skill 的名称或使用
/run-test-plan触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
**Added strict execution gates before running tests and evaluating results.**
- Verifies plan path, YAML validity, and evidence directory existence before starting tests.
- After setup (or with `--skip-setup`), checks that all required health checks/services are alive before any test runs.
- Adds validation that all artifacts from test steps (output, status, snapshots, evidence files) exist before pass/fail logic.
- Clarifies gate criteria for both legacy and modern setup formats.
- No functional changes to test execution, but reduces chances of tests running with missing prerequisites or incomplete state.
v1.0.0
Initial release of the run-test-plan skill.
- Executes a YAML-defined test plan: setup, health checks, and tests run sequentially.
- Stops on the first test failure and generates a detailed debug report with evidence and suggestions.
- Supports CLI, database, curl, and browser-agent test steps.
- Handles legacy and modern test plan formats.
- Outputs a summary test results table and preserves evidence for failures.
- Requires agent-browser skill for browser automation tests.
元数据
常见问题
Run Test Plan 是什么？
Execute YAML test plan, stop on first failure, output rich debug prompt. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 98 次。
如何安装 Run Test Plan？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install run-test-plan」即可一键安装，无需额外配置。
Run Test Plan 是免费的吗？
是的，Run Test Plan 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Run Test Plan 支持哪些平台？
Run Test Plan 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Run Test Plan？
由 Kevin Anderson（@anderskev）开发并维护，当前版本 v1.0.1。
推荐 Skills