Agent skill

Maa Pipeline Testing

Test and validate MaaFramework Pipeline JSON, recognition nodes, action nodes, CustomAction/CustomRecognition wiring, resource loading, and end-to-end task behavior.

AGPL-3.0Auto-check passedTesting & QA

Install Maa Pipeline Testing

skills CLI

$ npx skills add duorua/narutomobile --skill maa-pipeline-testing -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI

$ gh skill install duorua/narutomobile maa-pipeline-testing --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy

$ git clone --depth 1 https://github.com/duorua/narutomobile.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/maa-pipeline-testing .claude/skills/maa-pipeline-testing && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Paste into Claude Code

Install the "maa-pipeline-testing" agent skill from https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing into .claude/skills/maa-pipeline-testing/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "maa-pipeline-testing", then confirm the skill loads.

Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

$skill-installer

$skill-installer install https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing

Type this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.

skills CLI

$ npx skills add duorua/narutomobile --skill maa-pipeline-testing -a codex

Project install goes to .agents/skills/; add -g for ~/.codex/skills/.

GitHub CLI

$ gh skill install duorua/narutomobile maa-pipeline-testing --agent codex

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/duorua/narutomobile.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.agents/skills/maa-pipeline-testing .agents/skills/maa-pipeline-testing && rm -rf skills-src

Use ~/.agents/skills/ instead of .agents/skills for a personal install.

Codex skills documentation · loads skills from .agents/skills/

Paste into Codex

Install the "maa-pipeline-testing" agent skill from https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing into .agents/skills/maa-pipeline-testing/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "maa-pipeline-testing", then confirm the skill loads.

Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add duorua/narutomobile --skill maa-pipeline-testing -a cursor

Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.

GitHub CLI

$ gh skill install duorua/narutomobile maa-pipeline-testing --agent cursor

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/duorua/narutomobile.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.agents/skills/maa-pipeline-testing .cursor/skills/maa-pipeline-testing && rm -rf skills-src

Use ~/.cursor/skills/ instead of .cursor/skills for a personal install.

Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/

Paste into Cursor

Install the "maa-pipeline-testing" agent skill from https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing into .cursor/skills/maa-pipeline-testing/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "maa-pipeline-testing", then confirm the skill loads.

Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

gemini skills install

$ gemini skills install https://github.com/duorua/narutomobile.git --path .agents/skills/maa-pipeline-testing

--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.

skills CLI

$ npx skills add duorua/narutomobile --skill maa-pipeline-testing -a gemini-cli

Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.

GitHub CLI

$ gh skill install duorua/narutomobile maa-pipeline-testing --agent gemini-cli

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/duorua/narutomobile.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.agents/skills/maa-pipeline-testing .gemini/skills/maa-pipeline-testing && rm -rf skills-src

Use ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.

Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/

Paste into Gemini CLI

Install the "maa-pipeline-testing" agent skill from https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing into .gemini/skills/maa-pipeline-testing/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "maa-pipeline-testing", then confirm the skill loads.

Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

GitHub CLI

$ gh skill install duorua/narutomobile maa-pipeline-testing

Installs for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).

skills CLI

$ npx skills add duorua/narutomobile --skill maa-pipeline-testing -a github-copilot

Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.

Manual copy

$ git clone --depth 1 https://github.com/duorua/narutomobile.git skills-src && mkdir -p .github/skills && cp -r skills-src/.agents/skills/maa-pipeline-testing .github/skills/maa-pipeline-testing && rm -rf skills-src

Use ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.

GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/

Paste into GitHub Copilot

Install the "maa-pipeline-testing" agent skill from https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing into .github/skills/maa-pipeline-testing/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "maa-pipeline-testing", then confirm the skill loads.

GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add duorua/narutomobile --skill maa-pipeline-testing -a opencode

OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.

GitHub CLI

$ gh skill install duorua/narutomobile maa-pipeline-testing --agent opencode

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/duorua/narutomobile.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.agents/skills/maa-pipeline-testing .opencode/skills/maa-pipeline-testing && rm -rf skills-src

Use ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.

OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/

Paste into OpenCode

Install the "maa-pipeline-testing" agent skill from https://github.com/duorua/narutomobile/tree/main/.agents/skills/maa-pipeline-testing into .opencode/skills/maa-pipeline-testing/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "maa-pipeline-testing", then confirm the skill loads.

OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

Facts

Skill name: maa-pipeline-testing
GitHub stars: 338
Token cost: ~2.3k tokens
SKILL.md length: 467 words
Files: 2
Skills in repo: 11
Repo updated: Oct 9, 2026
First seen: Oct 7, 2026
Licence: AGPL-3.0

At a glance

Test and validate MaaFramework Pipeline JSON, recognition nodes, action nodes, CustomAction/CustomRecognition wiring, resource loading, and end-to-end task behavior.

Works in 5 steps: 生成测试 pipeline — 同一节点，多个 expand 变体 → 关键：expected 必须匹配当前资源实际显示文本；在 MaaGumballs… → run_pipeline 逐个跑，记录 score → …
Running runpipeline
SKILL.md covers 官方依据核对, 项目初始化接力, 概述 and 历史审查后的测试边界, plus 13 more sections
Calls rg and python

What it does

Maa Pipeline Testing is an agent skill from duorua/narutomobile. Test and validate MaaFramework Pipeline JSON, recognition nodes, action nodes, CustomAction/CustomRecognition wiring, resource loading, and end-to-end task behavior. Use when running runpipeline, checking OCR/TemplateMatch/ColorMatch ROI, validating context.runtask() results, auditing action: Custom or recognition: Custom registrations, or choosing the right resource-check command.

Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `agents/openai.yaml`).

It sits in Testing & QA. It works with Python. The licence is AGPL-3.0.

When your agent uses it

Running runpipeline
Checking OCR/TemplateMatch/ColorMatch ROI
Validating context.runtask() results
Auditing action: Custom

Example prompts

“/maa-pipeline-testing”

Requirements

Python 3

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

生成测试 pipeline — 同一节点，多个 expand 变体
关键：expected 必须匹配当前资源实际显示文本；在 MaaGumballs 中文资源里是 ["角色"]，不是英文 key ["Role"]
run_pipeline 逐个跑，记录 score
选 score ≥ 0.99 且 ROI 不重叠相邻元素的 expand
用 generate_node.py --expand X --overwrite 正式写入

What it can do on your machine

Read from SKILL.md and the folder at commit 16eacbf. It shows what the files ask for, not the result of running them.

Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
- rg
- python
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Maa Pipeline Testing loads about 2.3k tokens when it runs. Until then it costs about 104 tokens; SKILL.md has 467 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it: ~104
When it runs · the whole SKILL.md, loaded when a task matches: ~2.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from duorua/narutomobile at commit 16eacbf, republished under its AGPL-3.0 licence (© duorua). 467 words, ~2,317 tokens.

Download SKILL.mdSave it as .claude/skills/maa-pipeline-testing/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.

name: maa-pipeline-testing
description: Test and validate MaaFramework Pipeline JSON, recognition nodes, action nodes, CustomAction/CustomRecognition wiring, resource loading, and end-to-end task behavior. Use when running `run_pipeline`, checking OCR/TemplateMatch/ColorMatch ROI, validating `context.run_task()` results, auditing `action: Custom` or `recognition: Custom` registrations, or choosing the right resource-check command.

Pipeline Testing Skill

官方依据核对

测试命令、schema 字段或 MaaFramework API 行为以原始官方来源为准。涉及版本差异、binding 兼容性或语义变化时，通过 $maa-wiki 定位官方文档、schema 与源码后再下结论；测试通过不能替代对官方契约的核对。

如果用户要求验收一个尚未定义目标状态和完成条件的端到端自动化需求，先交给 $maa-workflow-build 建立任务契约；已有契约时，再用本 skill 为其中的验收条件收集测试证据。

项目初始化接力

测试前先查目标项目根目录的 basic_info.md。存在且包含第 0 节时，读取“0. Maa Skills 接力协议”和第 2/5/7/8/9/10 节，用 task/resource、公共返回节点、OCR/模板/ROI 和风险清单规划测试；随后以当前截图、当前源码和 TaskDetail 为准。优先把公共 Click 节点复制成临时 DoNothing 探针做初始化 smoke test，完成后调用 stop_pipeline 并删除临时文件。文件缺失或没有第 0 节时按本 skill 直接发现测试所需上下文；不得自动调用 $maa-project-init，只有用户明确要求初始化或刷新时才调用。源码更新更晚时视为缓存可能过期并以源码为准，不自动刷新或覆盖。

概述

测试 MaaFramework Pipeline JSON 中的 node，验证识别和操作是否正常工作。

历史审查后的测试边界

资源加载检查是 smoke test，不是端到端证明：python tools/ci/check_resource.py assets/resource/base 能证明资源包可加载，但不能证明 CustomAction/CustomRecognition 名称存在、参数路径正确、run_task() 分支判断正确。
Custom 映射必须单独查：所有 action: Custom / action.type = Custom 的 custom_action 都要能映射到 @AgentServer.custom_action(...)；所有 recognition.type = Custom 的 custom_recognition 都要能映射到 @AgentServer.custom_recognition(...)。
CustomRecognition 也要测：M9A 证明复杂 OCR/list/image 逻辑经常放在 CustomRecognition，不要只测 CustomAction。
高风险链路要复测稳定态：购买、消耗、战斗开始、继续挑战、结算确认等动作后，先识别稳定页面或完成态，再继续危险动作。
关闭节点不能继续调用：当 option 把可执行节点的 enable/enabled 设为 false 时，Python 必须先短路，不能再调用该节点的 run_recognition()/run_task()；同时验证关闭可选分支后，同级战斗/调查分支仍会继续识别。

Custom 映射快速检查

powershell

rg -n '"Custom"|custom_action|custom_recognition' assets/resource -g '*.json'
rg -n '@AgentServer\.custom_action|@AgentServer\.custom_recognition' agent -g '*.py'

人工对照时按名字精确匹配。资源检查通过但这里缺注册，运行时仍会失败。

核心流程

python

# 1. 连接设备
find_adb_device_list()
controller_id = connect_adb_device(device_name="xxx")
# 或
controller_id = connect_window(window_name="xxx")

# 2. 加载 pipeline
load_pipeline(pipeline_path="<pipeline_json>")
RESOURCE_PATH = "<resource_base_path>"

# 3. 逐个测试
for node_name in pipeline_nodes:
    result = run_pipeline(
        controller_id=controller_id,
        pipeline_path=pipeline_path,
        entry=node_name,
        resource_path=RESOURCE_PATH
    )
    # succeeded + all_results 有内容 = 识别成功
    # 详见下方"结果判断"小节

结果判断

status	all_results	含义
`succeeded`	有内容	✅ 识别成功
`succeeded`	空	❌ 识别失败
`failed`	—	❌ 节点未触发 / 识别超时

返回结构：

json

{
  "status": "succeeded",
  "nodes": [{
    "name": "node_name",
    "recognition": {
      "all_results": [{"box": [x, y, w, h], "score": 0.99, "text": "..."}]
    }
  }]
}

`run_task()` 结果判断 ⚠️

MCP 的 run_pipeline 单节点结果和 Python CustomAction 里的 context.run_task() 结果不是同一个使用场景。context.run_task() 的 result.nodes 可能包含:

真正命中并执行完成的节点
next 列表里被尝试过但失败的节点
调试面板显示红叉的节点

因此不要写:

python

# ❌ 错：有 nodes 不代表目标节点命中
result = context.run_task("AutoSky_CloneDied")
if result and result.nodes:
    handle_clone_loss()

应该检查节点是否真的完成或识别命中:

python

def task_result_has_hit(result, names: set[str]) -> bool:
    if not result or not result.nodes:
        return False
    for node in result.nodes:
        if getattr(node, "name", None) not in names:
            continue
        if getattr(node, "completed", False):
            return True
        recognition = getattr(node, "recognition", None)
        if recognition and getattr(recognition, "hit", False):
            return True
    return False

实战信号：调试面板里目标节点是红叉，但日志却进入了成功分支，基本就是把"节点出现过"误当成"节点命中"。

对于会回到稳定页面的流程，建议先检测稳定状态:

python

if task_result_has_hit(result, {"AutoSky_CheckExplorationInfo"}):
    return "done"

再跑危险兜底动作（取消结算、购买确认、继续挑战等），避免页面已经回稳后误触发。

跨测试导航 ⚠️

Click 节点会让页面跳转，破坏后续测试初始状态。必须用 BackButton_500ms 返回：

python

# 测试节点
run_pipeline(..., entry="NodeA", ...)

# 必须返回！不要 click_key(4) 模拟 ESC（可能进错页面）
run_pipeline(..., entry="BackButton_500ms", ...)

# 测下一个
run_pipeline(..., entry="NodeB", ...)

BackButton_500ms 在 main_ui.json 里，DirectHit 识别返回箭头，定位精确可靠。

ROI Sweep 测试方法

OCR 失败时不要换 TemplateMatch！先用 sweep 找 sweet spot。

操作流程：

生成测试 pipeline — 同一节点，多个 expand 变体
关键：expected 必须匹配当前资源实际显示文本；在 MaaGumballs 中文资源里是 ["角色"]，不是英文 key ["Role"]
run_pipeline 逐个跑，记录 score
选 score ≥ 0.99 且 ROI 不重叠相邻元素的 expand
用 generate_node.py --expand X --overwrite 正式写入

实战矩阵（5 节点实测）：

节点	e0	e20	e50	最佳
角色	✅	✅ 0.9997	✅	e20
编队	✅	✅ 0.998	✅	e20
城堡	✅	❌	—	e3（仅 0-15）
佣兵团	❌	✅	✅	e20
前往群岛	✅	✅	✅	e20

观察：

OCR 引擎非确定性：同一 ROI 不同次测试结果可能不同
timeout: 2000 期间 OCR 会重试多次，所以单次失败不一定是真失败
特殊节点（如"城堡"）需小 ROI，因为大 ROI 包含邻近 UI 元素会干扰 OCR

资源保护 ⚠️

测试时绝对不要点击这些按钮（消耗资源）：

升级建筑、神殿升级
供奉、祭拜先祖
购买物品、商城购买
确认战斗开始
任何有资源消耗的确认按钮

误入后处理：尝试 BackButton_500ms 或 ESC → 切换 tab 刷新。

常见问题

现象	排查
节点超时失败	截屏看实际界面 → 调整 `expected` 文本或 `roi`
OCR 拆分多字（如"角色"→"电"+"色"）	缩小 ROI 避开干扰
误识别相邻文字	缩小 ROI 限制范围
点击位置不准	实际点击取 box 中心点：`x + w/2, y + h/2`
可滚动 UI 单视图测不全	一个截图里只能看到 5-6 个元素，需要 swipe 多次分别测；详见下方
Click + BackButton 跳错页面	Click 子页后按 BackButton 可能回到大地图/外层 UI，不是预期父页面；需主动 navigate 回来
部分可见元素识别	边界处只露 25px 也能 OCR 找到（score 偏低 0.8x），用大 ROI 覆盖即可

Show full SKILL.md (173 more words)Show less

可滚动 UI 多视图测试

单次截图只能测当前可见的 5-6 个元素。要测全列表（如城堡 10 个建筑）必须多次 swipe：

1. 进入目标页（如城堡顶部）
2. run_pipeline 测当前可见的 N 个节点
3. 每个 click 后 BackButton 返回
4. swipe 滚动一屏
5. 再 run_pipeline 测新可见的节点
6. 重复直到列表测完

典型案例：城堡建筑（10 节点）

顶部：城堡管理/市场/铁匠铺/炼金工坊/训练所
中段：城堡主厅/神殿/家族
底部：藏馆/庄园

关键：所有节点共用同一大 ROI [x, top_y, w, full_h]，不要给每个节点算不同 ROI（滚动后位置变了，原 ROI 失效）。

卡住时截图查看

症状：节点超时、OCR 找不到、行为异常、识别 score 突然下降

解决：调 screencap(controller_id) 看实际屏幕状态：

当前在哪个页面？（是否已跳到其他页面）
目标文字是否被遮挡/弹窗挡住？
屏幕是否在加载/转圈？
滚动位置对吗？

配合 ocr() 看识别结果交叉验证。不要盲调 ROI/expected，先看实际屏幕。

跨测试导航 ⚠️（Click 节点路径风险）

Click 节点会让页面跳转，破坏后续测试初始状态。必须用 BackButton_500ms 返回：

python

# 测试节点
run_pipeline(..., entry="NodeA", ...)

# 必须返回！不要 click_key(4) 模拟 ESC（可能进错页面）
run_pipeline(..., entry="BackButton_500ms", ...)

注意：BackButton 不一定回到你期望的父页面！

例：城堡管理 → BackButton → 实际回到 大地图（不是城堡）
例：训练所 → BackButton → 回到城堡（正常）

测试前先确认 BackButton 路径，或者用 ocr() 检查当前页面状态再继续。

⚠️ MCP `start_agent=true` 失败时的 Fallback 方案

症状:run_pipeline(..., start_agent=true) 报

Agent 启动失败: Agent 启动失败 (identifier=12345): connect() 返回 False

根因(常见):

MaaFramework 的 agent 子进程(socket 进程)未在预期时间窗口内完成 MaaAgentServerStartUp
Socket 路径冲突 / 权限问题
MaaFramework 内部 MaaAgentServerNotImpl(这个错误明确说"用 MaaFramework,不要用 MaaAgentServer")

Fallback 1:单节点测试(推荐):

用 start_agent=false + 指定 entry 测单个节点(只测 OCR/Template 识别,跳过 CustomAction 链):

python

# 不开 agent,直接测某个节点
result = run_pipeline(
    controller_id=cid,
    pipeline_path="<project-root>/assets/resource/base/pipeline/example.json",
    entry="AutoSky_BagConfig",   # 任意节点名
    resource_path="<project-root>/assets/resource/base",
    start_agent=False,          # ← 关键:false
)

适用场景:测 OCR / TemplateMatch / ColorMatch 等识别节点的 ROI 和 expected 文本是否正确。

Fallback 2:手动启动 agent 后 MCP 通过 socket 连接:

如果单节点测试不够(需测完整任务链),手动跑 agent 进程,然后用 connect_adb_device / connect_window 让 MCP 复用。

python

# 步骤 1:手动启动 agent
import subprocess
agent_proc = subprocess.Popen(
    ["python", "agent/main.py", "test_socket"],
    cwd=".",
)

# 步骤 2:让 MCP 连接(需要 MCP 内部 socket ID 匹配)
# 实际中,MCP 启动 agent 会自动起 socket,如果手动起需要 MCP 端配对
# 通常这一步是 MCP 内部行为,我们无法直接控制

Fallback 3:放弃 MCP,直接用 MaaFramework Python API:

python

import sys
sys.path.insert(0, "<project-root>")
from maa.agent.agent_server import AgentServer
from maa.toolkit import Toolkit
from maa.resource import Resource
from maa.controller import AdbController
from maa.tasker import Tasker

Toolkit.init_option("./")
res = Resource(); res.post_bundle("./assets/resource/base").wait()
ctrl = AdbController(adb_path="adb", address="127.0.0.1:16384"); ctrl.post_connection().wait()
tasker = Tasker(); tasker.bind(res, ctrl)

import action.sky
autosky = action.sky.AutoSky()
result = autosky.run(tasker.context, type('X', (), {'custom_action_param': '{}'})())

何时用哪种:

Fallback 1(单节点):调试 ROI / expected / TemplateMatch,快速迭代 → 首选
Fallback 3(直接 API):测完整 task 链,真实环境验证 → MCP 不可用时用
Fallback 2(手动起 agent):通常不需要,MCP 会自动处理

测试模板

markdown

## <文件名> 测试记录

日期: 2026-06-05
设备: xxx
controller_id: xxx

### Node 列表
- [ ] node_name_1
- [ ] node_name_2

### 测试进度
| Node | 结果 | 备注 |
|------|------|------|
| xxx  | ✅   | score=0.999 |
| xxx  | ❌   | 原因xxx |

Reference（详见 maa-pipeline-guide）

ROI / box / target 概念
识别类型：DirectHit / OCR / TemplateMatch / FeatureMatch / ColorMatch / And / Or
动作类型：Click / LongPress / Swipe / Scroll / InputText / ClickKey
节点生命周期：pre_wait_freezes → pre_delay → action → post_wait_freezes → post_delay → 截图识别 next

Files

SKILL.md and 1 other file in .agents/skills/maa-pipeline-testing of duorua/narutomobile.

SKILL.md
agents/openai.yaml

Open the folder on GitHubat commit 16eacbf

Compare with similar skills

Maa Pipeline Testing next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Maa Pipeline Testing compared with similar skills
Skill	Stars	Used in	Tokens	Auto-check	Licence	Repo updated
Maa Pipeline Testing this skillduorua/narutomobile	338	—	~2.3k	Automated safety check: Pass	AGPL-3.0	today
Web Application Testinganthropics/skills	180k	51 repos	~966	Automated safety check: Pass	Apache-2.0	today
OpenHarness End-to-End EvalsHKUDS/OpenHarness	16k	1 repos	~2.1k	Automated safety check: Notes	MIT	4 mo ago
Adk Verify Snippetsgoogle/adk-python	22k	—	~1.4k	Automated safety check: Pass	Apache-2.0	today
Apple Container Test RunnerRustPython/RustPython	22k	—	~467	Automated safety check: Pass	MIT	today
JSON Repair Docs Demo Local Testmangiucugna/json_repair	5.1k	—	~549	Automated safety check: Pass	MIT	yesterday

Similar skills

Web Application Testing
anthropics/skills
Official
Tests local web applications with Python Playwright scripts, checking frontend behavior, capturing screenshots and reading browser console logs.
180k GitHub starsUsed in 51 repos~966 tokens
Testing & QAAuto-check passed
OpenHarness End-to-End Evals
HKUDS/OpenHarness
Validates OpenHarness features by running real multi-turn agent loops with live LLM calls against an unfamiliar codebase, checking actual tool execution.
16k GitHub starsUsed in 1 repo~2.1k tokens
Testing & QAAuto-check: notes
Adk Verify Snippets
google/adk-python
Official
Checks that every Python code block in a Markdown file actually compiles and runs, by extracting each block to a temporary file, executing it in an isolated subprocess, and writing a pass/fail…
22k GitHub stars~1.4k tokensUpdated today
Testing & QAAuto-check passed
Apple Container Test Runner
RustPython/RustPython
Runs RustPython tests inside a Linux container built with Apple's container CLI, so macOS users can compare Linux results with their local ones.
22k GitHub stars~467 tokensUpdated today
Testing & QAAuto-check passed
JSON Repair Docs Demo Local Test
mangiucugna/json_repair
Runs the json_repair docs demo against a local Flask API and static server, so changes to docs/app.py or the docs UI are checked end to end before publishing.
5.1k GitHub stars~549 tokensUpdated yesterday
Testing & QAAuto-check passed
Wobbling Ty Constraint Order
astral-sh/ruff
Official
A skill your agent uses when a user asks to wobble ty constraint ordering, check constraint-set or TDD ordering determinism, test reversed constraint/typevar IDs, or investigate nondeterministic ty…
50k GitHub stars~838 tokensUpdated today
Testing & QAAuto-check passed

Questions about Maa Pipeline Testing

What does Maa Pipeline Testing do?

Test and validate MaaFramework Pipeline JSON, recognition nodes, action nodes, CustomAction/CustomRecognition wiring, resource loading, and end-to-end task behavior. Maa Pipeline Testing is an agent skill from duorua/narutomobile. Test and validate MaaFramework Pipeline JSON, recognition nodes, action nodes, CustomAction/CustomRecognition wiring, resource loading, and end-to-end task behavior.

When should I use Maa Pipeline Testing?

Maa Pipeline Testing fits situations like: running runpipeline; checking OCR/TemplateMatch/ColorMatch ROI; validating context.runtask() results; auditing action: Custom.

How do I install Maa Pipeline Testing in Claude Code?

Run `npx skills add duorua/narutomobile --skill maa-pipeline-testing -a claude-code`. Or copy the skill folder (.agents/skills/maa-pipeline-testing in duorua/narutomobile) into .claude/skills/maa-pipeline-testing in your project. Claude Code loads it when a task matches its description.

How do I install Maa Pipeline Testing in Codex?

Run `npx skills add duorua/narutomobile --skill maa-pipeline-testing -a codex`. Or copy the skill folder (.agents/skills/maa-pipeline-testing in duorua/narutomobile) into .agents/skills/maa-pipeline-testing in your project. Codex loads it when a task matches its description.

Can I use Maa Pipeline Testing in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add duorua/narutomobile --skill maa-pipeline-testing -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/maa-pipeline-testing, .gemini/skills/maa-pipeline-testing, .github/skills/maa-pipeline-testing and .opencode/skills/maa-pipeline-testing in your project.

What does Maa Pipeline Testing need to run?

Going by SKILL.md and its folder, Maa Pipeline Testing needs the command-line tools its instructions call (rg and python). Our summary lists: Python 3.

Does Maa Pipeline Testing access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Maa Pipeline Testing safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Maa Pipeline Testing use?

Maa Pipeline Testing is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Maa Pipeline Testing use?

About 2.3k tokens (SKILL.md is roughly 9.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Maa Pipeline Testing?

Skills that share tags, products or a category with Maa Pipeline Testing: Web Application Testing (anthropics/skills, 180k stars), OpenHarness End-to-End Evals (HKUDS/OpenHarness, 16k stars), Adk Verify Snippets (google/adk-python, 22k stars) and Apple Container Test Runner (RustPython/RustPython, 22k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Maa Pipeline Testing?

duorua (a GitHub user) maintains it in duorua/narutomobile, which has 338 GitHub stars. The repository holds 11 skills in this directory. The repository was last updated on October 9, 2026.

Source: duorua/narutomobile on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.

Install Maa Pipeline Testing

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

Pipeline Testing Skill

官方依据核对

项目初始化接力

概述

历史审查后的测试边界

Custom 映射快速检查

核心流程

结果判断

run_task() 结果判断 ⚠️

跨测试导航 ⚠️

ROI Sweep 测试方法

资源保护 ⚠️

常见问题

可滚动 UI 多视图测试

卡住时截图查看

跨测试导航 ⚠️（Click 节点路径风险）

⚠️ MCP start_agent=true 失败时的 Fallback 方案

测试模板

Reference（详见 maa-pipeline-guide）

Files

Compare with similar skills

Similar skills

More from duorua/narutomobile

Related

Questions about Maa Pipeline Testing

`run_task()` 结果判断 ⚠️

⚠️ MCP `start_agent=true` 失败时的 Fallback 方案