API审查
- 作者仓库星标 0
- 作者更新于 实时读取
- 作者仓库 skills-registry
- 领域
- AI 智能
- 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
- @tomevault-io · 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
- 需要 · Vendor-specific
- 兼容的系统
- 未声明(默认跨平台)
- 底层运行要求
- 无特殊要求
- 文件与系统权限
-
- 只读
- 允许写入 / 修改
- Shell 执行
- 网络行为
- 允许外网请求
- 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: coding-cli
description: 将编码工作委托给 Manager 持有的 AI CLI 工具(Claude Code / Gemini CLI / qodercli)执行 Use when this capability i…
category: AI 智能
runtime: 无特殊运行时
---
# coding-cli 输出预览
## PART A: 任务判断
- 适用问题:提示词、Agent 工作流、模型评估或自动化推理。
- 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“When to Use This Skill / Step-by-Step Workflow / 1. Prepare the Workspace”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提示词、Agent 工作流、模型评估或自动化推理,并标出输入、限制和预期结果。
- **02** 执行计划:优先按“When to Use This Skill / Step-by-Step Workflow / 1. Prepare the Workspace”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、会按任务需要访问外部网络、需要准备 Vendor-specific 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;会按任务需要访问外部网络;需要准备 Vendor-specific 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。 原文出现了 `/path`、`/root` 这类斜杠命令;如果你的 Agent 支持命令触发,优先用命令开场,再补充目标和边界。
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“When to Use This Skill / Step-by-Step Workflow / 1. Prepare the Workspace”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: coding-cli
description: 将编码工作委托给 Manager 持有的 AI CLI 工具(Claude Code / Gemini CLI / qodercli)执行 Use when this capability i…
category: AI 智能
source: tomevault-io/skills-registry
---
# coding-cli
## 什么时候使用
- 把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能 适合处理AI Agent、提示词、模型评估与自动化推理,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…
- 面向提示词、Agent 工作流、模型评估或自动化推理,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「When to Use This Skill / Step-by-Step Workflow / 1. Prepare the Workspace」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;会按任务需要访问外部网络;需要准备 Vendor-specific 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 "coding-cli" {
输入层 -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
上下文层 -> When to Use This Skill / Step-by-Step Workflow / 1. Prepare the Workspace
规则层 -> SKILL.md 触发条件 / 执行顺序 / 输出格式
运行层 -> 无特殊运行时 | 读取文件、写入/修改文件、执行终端命令 | 会按任务需要访问外部网络
安全层 -> 需要准备 Vendor-specific 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
} Coding CLI Delegation
When the Manager assigns you a task with a "## Coding CLI Mode" section in spec.md, you do not write the code yourself. Instead, you:
- Understand the task deeply
- Prepare the workspace
- Generate a precise coding prompt
- Delegate execution to the Manager's CLI tool
- Review the result
When to Use This Skill
Check spec.md for the task. If it contains ## Coding CLI Mode, use this skill for all code changes.
Step-by-Step Workflow
1. Prepare the Workspace
Set up the workspace directory under the shared filesystem:
workspace="/root/hiclaw-fs/shared/tasks/{task-id}/workspace"
mkdir -p "$workspace"
# Clone a repo (example)
git clone <repo-url> "$workspace"
# Or copy existing code
cp -r /path/to/source "$workspace/"
Constraint: The workspace path must be under /root/hiclaw-fs/. The Manager accesses the same path via MinIO mirror.
2. Push Workspace to MinIO
Before sending the coding-request, push all workspace files so the Manager can access them:
mc mirror "/root/hiclaw-fs/shared/tasks/{task-id}/workspace/" \
${HICLAW_STORAGE_PREFIX}/shared/tasks/{task-id}/workspace/
2b. Check for Processing Marker
Before modifying the workspace or sending a coding-request, check if the task directory is being processed:
# Sync latest state from MinIO
mc mirror "${HICLAW_STORAGE_PREFIX}/shared/tasks/{task-id}/" \
"/root/hiclaw-fs/shared/tasks/{task-id}/"
# Check for processing marker
if [ -f "/root/hiclaw-fs/shared/tasks/{task-id}/.processing" ]; then
echo "Task directory is being processed. Wait for manager to complete."
# Do NOT send coding-request yet; wait and retry
fi
If a .processing marker exists, wait for the Manager to complete their operation before sending your request.
3. Generate a High-Quality Prompt
A good prompt includes:
- Target files: exact paths and relevant line numbers
- Specific changes: describe what to change, not just why
- Context: existing code structure, dependencies, interfaces
- Acceptance criteria: how to verify the change is correct
- Constraints: languages, frameworks, style conventions, things NOT to change
Example of a good prompt:
In the file `src/server/handlers/auth.go`, implement the `RefreshToken` function (currently a stub at line 142).
Requirements:
- Validate the incoming refresh_token from the request body (field name: "refresh_token")
- Look up the token in the database using `db.FindRefreshToken(ctx, token)` (already imported)
- If valid, generate a new access token with `auth.GenerateAccessToken(userID)` and return it as JSON: {"access_token": "<token>", "expires_in": 3600}
- If invalid or expired, return HTTP 401 with body: {"error": "invalid_refresh_token"}
- Follow the existing error handling pattern used in `LoginHandler` (line 89)
Do not change any other files.
Signs of a weak prompt (avoid):
- "Fix the auth system" (too vague)
- "Improve performance" (no specific target)
- "Add tests" (no specification of what to test or where)
4. Send coding-request: to Manager
Send in your Worker Room (or Project Room, wherever the task was assigned):
@manager:DOMAIN task-{task-id} coding-request:
workspace: /root/hiclaw-fs/shared/tasks/{task-id}/workspace
---PROMPT---
{your detailed coding prompt here}
---END---
Note: workspace can be any subdirectory under /root/hiclaw-fs/, e.g. a cloned git repo:
workspace: /root/hiclaw-fs/shared/tasks/{task-id}/workspace/my-repo
5. Wait for Manager's Response
The Manager will run the CLI tool and respond with either:
Success — coding-result:
@{your-name}:DOMAIN task-{task-id} coding-result:
CLI 工具已完成编码。请同步工作目录并 review 变更...
Failure — coding-failed:
@{your-name}:DOMAIN task-{task-id} coding-failed:
CLI 工具执行失败...你生成的提示词已保存于:/root/hiclaw-fs/shared/tasks/{task-id}/coding-prompts/
6a. On coding-result:
# Sync changes from MinIO
hiclaw-sync
# Review what changed
cd /root/hiclaw-fs/shared/tasks/{task-id}/workspace
git diff # if it's a git repo
# or: check coding-cli-logs/ for CLI output
Review the changes:
- Verify they match the task requirements
- Check for obvious errors or unintended modifications
- Run tests if applicable
Report to Manager:
@manager:DOMAIN task-{task-id} completed:
Changes reviewed and verified. {Brief summary of what was implemented.}
6b. On coding-failed:
Implement the coding task yourself using your normal approach. When done, report:
@manager:DOMAIN task-{task-id} completed:
Implemented manually (CLI delegation failed). {Brief summary.}
Multiple Rounds
If one CLI delegation doesn't fully complete the task (e.g., the first run fixed the bug but tests still fail), you can send another coding-request: with a follow-up prompt. Sync the workspace first to get the latest state before generating the next prompt.
Tips for Better Prompts
- Be surgical: "change line 47 from X to Y" beats "fix the login function"
- Provide imports: if the change requires new imports, specify them
- Show the pattern: reference existing code in the same file as the style guide
- Limit scope: tell the CLI "only modify file X, do not change tests or other files"
- Verify command: include how to check correctness, e.g. "run
go test ./...to verify"
Source: agentscope-ai/HiClaw — distributed by TomeVault.
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