一键导入
coding-plan
Read before planning, declare scope before implementing. Prohibit out-of-scope changes and full rewrites.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Read before planning, declare scope before implementing. Prohibit out-of-scope changes and full rewrites.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
Translate a web page while preserving its original DOM structure, CSS, and images. Generates a self-contained bilingual preview HTML with toggle controls. Use this skill whenever the user shares a URL and asks for translation, translated preview, bilingual view, or when the channel topic instructs to use translate-preview. Do NOT write your own translation scripts — always use this skill's reference files.
Job scheduling. Manage recurring tasks using MCP tools.
Generate a prioritized daily briefing with calendar, email, tasks, and yesterday's activity. Designed for cron execution. Trigger with "morning briefing", "daily brief", "start my day".
Self-management via geminiclaw_admin MCP tool. Config, skills, upgrade, sessions.
Proactively explore the web for topics matching the user's interests and share findings conversationally. Designed for cron execution with Flash model.
| name | coding-plan |
| description | Read before planning, declare scope before implementing. Prohibit out-of-scope changes and full rewrites. |
| enabled | true |
A skill applied to all coding tasks that modify existing files. Adapted from Claude Code's EnterPlanMode + TaskCreate approach for the Gemini agent.
Use this skill when the following apply:
Principle: Never edit or propose changes without reading first.
Declare the following to the user before execution:
✅ What will be changed
- {file}: {change description} — Done when: {verifiable criteria}
- ...
❌ Out of scope (will not be changed)
- {file/feature}: Outside the scope of this request
⚠️ Irreversible operations (if any)
- {operation}: {impact scope} — Will confirm before executing
If the scope is ambiguous, ask questions before declaring:
Use skills/todo-tracker/SKILL.md to record in TODO.md:
Example:
- [ ] Add {function} to src/foo.ts <!-- Done when: no tsc errors, tests pass -->
- [ ] Add tests to tests/foo.test.ts <!-- Done when: npm test passes -->
Prefer small, targeted edits to existing files (full rewrites are prohibited).
Prohibited patterns:
Check off each step in TODO.md upon completion.
Confirm with the user before executing irreversible operations (file deletion, dependency removal, breaking API changes).
## Done: {task name}
### What was changed
- `{file}`: {change description}
### Verification results
- {test results / build results}
### Out of scope (not changed)
- {reason}
| Prohibited | Reason |
|---|---|
| Editing without reading first | Breaks existing patterns / creates duplication |
| Adding improvements beyond the request scope | Introduces changes the user did not intend |
| Rewriting an entire file at once | Makes diffs hard to follow / risks deleting existing logic |
| Executing irreversible operations without confirmation | Causes unexpected loss for the user |
| Brute-forcing without stopping to analyze | Hides root causes and increases technical debt |