Skills Plugins MCP Prompt Model 博客 我的中心

local-frontend-check

Smoke-test or verify UI behaviour on the local Jarvis Registry frontend running at http://localhost/gateway. Use for manual regression checks, bug-fix verification, and end-to-end confirmation of specific flows without running the automated test suite.

DeepseekModel 官方收录技能 质量 优秀 · 90 v1.0.0

获取

https://deepseekmodel.com/api/download.php?id=ascending-llc-jarvis-registry-github-skills-local-frontend-check-skill-md&format=skill
下载 .skill 标准格式,含 system_prompt 与 model_config,导入任意 Agent 框架即可使用
.skill 文件中 system_prompt 字段的实际内容。
name local-frontend-check description Smoke-test or verify UI behaviour on the local Jarvis Registry frontend running at http://localhost/gateway. Use for manual regression checks, bug-fix verification, and end-to-end confirmation of specific flows without running the automated test suite. allowed-tools Bash(playwright-cli:*) Bash(docker:*) Local Frontend Check Skill Fixed Configuration Setting Value Base URL http://localhost/gateway Browser session Named jarvis — keeps a stable, reusable identity across runs Launch flags --persistent --headed — profile saved to disk, browser window visible Backend log container <cwd-basename>-registry-1 — where <cwd-basename> is the basename of your current working directory (e.g. if CWD is /projects/jarvis-registry , use jarvis-registry-registry-1 ) Workflow Follow these phases in order every time this skill is invoked. Do not skip phases. Phase 1 — Open the browser playwright-cli -s=jarvis open http://localhost/gateway --persistent --headed Then take a snapshot to read the page's current state: playwright-cli -s=jarvis snapshot Phase 2 — Handle authentication Inspect the snapshot. If the app loads directly (no login form, no auth redirect visible): the persistent profile is still valid. Skip to Phase 3. If a login page or auth redirect is visible : the session has expired or this is a first run. Do the following: Tell the user: "The browser window is open and showing a login page. Please log in in the headed browser window, then confirm here when done." Use AskUserQuestion to wait for the user's confirmation before continuing. Take a second snapshot and verify the app has loaded (URL is no longer the login page, and the main app UI is visible). If still on the login page, tell the user and repeat. Because --persistent saves the profile to disk, subsequent runs will pick up the stored session automatically and this phase will be skipped. Phase 3 — Orient and navigate Read the user's description of what to check. Navigate to the relevant section of the UI: Use playwright-cli -s=jarvis goto <url> for direct navigation. Use playwright-cli -s=jarvis snapshot --depth=3 to get a lightweight view of a complex page. Use playwright-cli -s=jarvis click <ref> to follow navigation links or open panels. Take snapshots as needed to orient yourself before acting. Phase 4 — Perform the described operations Execute the operations step by step. After each significant action (form submit, dialog confirm, delete button, save button), take a snapshot immediately to observe the result. Note what you see: Success signals : confirmation toast/banner, resource list updated, no error visible, HTTP response in UI shows 2xx. Failure signals : error toast, error dialog, HTTP 4xx/5xx error text, UI stuck in loading state. Capture a screenshot at any notable point (success or failure): playwright-cli -s=jarvis screenshot Phase 5 — Read backend logs After triggering the key operation(s), read recent logs from the backend container: docker logs <cwd-basename>-registry-1 -- tail =200 2>&1 # Replace <cwd-basename> with the basename of your CWD (e.g. `jarvis-registry` if CWD is `/projects/jarvis-registry`) Look for: Python tracebacks or ERROR -level log lines near the time of the operation. HTTP route log lines showing the request method, path, and status code. Any RuntimeError , ValueError , or DuplicateKeyError messages. If the log output is noisy, filter for the operation's timeframe by scanning the last N lines around when you performed the action. Phase 6 — Report findings Summarise clearly: Operations performed — what you did and in what order. UI result — what the browser showed after each operation (include snapshot filenames or key text). Log evidence — relevant log lines (errors, warnings, or confirmation of a clean request/response). Verdict — PASS or FAIL with a one-sentence reason. Notes Do not close the browser session at the end unless the user asks. Leaving it open avoids re-login overhead on follow-up checks. If the app is unreachable ( ERR_CONNECTION_REFUSED ), tell the user that the services may not be running and suggest docker compose up -d . Use playwright-cli -s=jarvis snapshot <ref> to inspect a specific element more closely when the full-page snapshot is too large. If an operation opens a confirmation dialog, handle it explicitly with playwright-cli -s=jarvis click <confirm-button-ref> rather than assuming it auto-dismisses.
Agent 识别该技能的关键词,点击任意一个即可复制。

该技能未提供触发词。

下载的 .skill 包内含以下字段。
字段 说明
format格式标识(skill/v1)
skill_id技能唯一 ID
name技能名称
version版本号
description技能描述
category所属分类(数组)
trigger_words触发词列表
tags标签列表
source来源标识
source_url来源链接(本页地址)
exported_at导出时间(每次下载生成)
system_prompt系统提示词正文
model_config模型参数:provider / model / temperature / max_tokens / top_p
examples示例
install_guide各平台导入说明(Coze / Dify / Claude / 自定义框架)
同一份技能可按不同平台格式导出。
.skill 标准格式,含 system_prompt 与 model_config,导入任意 Agent 框架即可使用 下载
.skillpro 增强格式,额外含脚本 / 工具 / 依赖 / 钩子占位 下载
.json 纯 JSON 导出,只含 system_prompt 与模型参数 下载
Coze 带 frontmatter 的 Markdown,Coze 平台导入用 下载
Dify Dify DSL,创建应用后直接导入 下载

每日精选 Skill 推荐,免费送到你邮箱

输入邮箱,每天接收一个精选 AI Agent 技能推荐。完全免费,持续更新。

提交后我们会发送一封确认邮件,点击邮件里的链接才会开始收信。

完全免费,取消任意时间。我们不会发送垃圾邮件。