AI生成

AI生成技能 - 负责广告创意的AI生成处理

Security Scan

The skill's code, instructions, and requirements are coherent with an ad-generation helper that simulates generation locally; there are no requests for credentials, external downloads, or unexpected system access.

Purpose & Capability

Name/description (ad creative generation) match the included code: an in-memory task queue that creates, lists, cancels, scores, deduplicates, and selects generated materials. No unrelated credentials, binaries, or external services are requested.

Instruction Scope

SKILL.md and index.ts are largely consistent about capabilities, but there are small mismatches: SKILL.md examples show api.executeAction with names like 'ai-generation.create' and event names 'ai-generation.completed' while the code registers an 'ai' command namespace and emits events 'generation.batch_completed' and 'generation.completed'. Also SKILL.md says 'TypeScript compiled after use' but no install/build step is provided by the package; runtime depends on the platform to execute the TypeScript (or a compiled JS). These are operational inconsistencies (not malicious) that could cause integration/runtime confusion.

Install Mechanism

No install spec or external downloads. The package includes TypeScript source and a build script, with no external dependencies. Low installation risk — nothing is fetched from external URLs or installed to unusual locations.

Credentials

The skill requests no environment variables, secrets, or config paths. It does read api.config (platform-provided configuration) which is expected for configurable options like defaultModel/maxConcurrent.

Persistence & Privilege

always is false and the skill does not request permanent system-wide privileges. The skill stores tasks in-memory (no persistent storage access) and does not modify other skills or global agent settings.

Assessment

This skill appears to be a local/in-memory ad-generation helper and doesn't request any secrets or external network access. Before installing, check the following: (1) confirm the event and action names your agent expects — SKILL.md examples and the code use different names, which may break integration; (2) ensure your runtime will compile/run the provided TypeScript (package.json has a build script but no install hook); (3) be aware tasks are kept in memory only (no persistence), so results will not survive a restart unless your platform persists them; and (4) if you plan to connect the skill to real AI/model endpoints, verify how credentials or network calls will be added and reviewed. Otherwise there are no security red flags in the provided files.

AI Generation - 广告创意AI生成

负责广告创意的AI生成处理，包括文案生成、图像生成、视频生成等功能。

Setup

无需额外依赖，TypeScript编译后使用。

When to Use

需要生成广告创意内容时使用：

• 广告文案生成
• 营销图片生成
• 视频素材制作
• 创意素材批量生成

Architecture

ai-generation/
├── index.ts          # 主入口，处理生成任务
└── package.json      # 依赖配置

Core Commands

创建生成任务

await api.executeAction('ai-generation.create', {
  demandId?: string,     // 关联的需求ID
  prompt: string,        // 生成提示词
  model?: string,        // 使用的模型（默认flux）
  params?: Record<string, any>,  // 额外参数
  count: number          // 生成数量
});

查询生成任务

await api.executeAction('ai-generation.get', {
  taskId: string         // 任务ID
});

取消生成任务

await api.executeAction('ai-generation.cancel', {
  taskId: string         // 任务ID
});

响应事件

• ai-generation.completed - 生成完成
• ai-generation.failed - 生成失败
• ai-generation.progress - 进度更新

Configuration

配置项（通过OpenClaw配置系统）：

• defaultModel: 默认使用的AI模型
• maxConcurrent: 最大并发数
• timeout: 超时时间（秒）