📦 Openclaw Meta Ads
v1.0.0共享的 Meta Ads 技能，供 OpenClaw agents 使用。分析 Meta（Facebook 和 Instagram）广告账户、推广活动、广告组、广告、成效洞察及线索表单数据……
0· 347·0 当前·0 累计
by 下载技能包
最后更新
2026/3/15
安全扫描
OpenClaw
安全
medium confidence这是一个以文档为先的 Meta Ads 技能，其声明的行为（先读后分析，写入前需显式批准）与其说明一致；使用前仅需检查少量打包/可见性问题。
评估建议
This skill appears coherent and safe as a docs-first Meta Ads auditing tool. Before installing: 1) confirm you will supply Meta tokens (e.g., bearer token / META_ADS_API_KEY) via your environment/secure credential injection and limit scopes to ads_read unless you explicitly approve writes; 2) verify the missing scripts referenced in README are intentionally absent — if scripts are later added, review them before use; 3) be cautious with the browser-fallback path — ensure the browser context is i...详细分析 ▾
✓ 用途与能力
Name and description align with the content: the skill is a shared Meta Ads guidance and audit/reporting workbook. The references describe typical tokens (e.g., META_ADS_API_KEY) and account IDs, which are appropriate for the purpose. One minor inconsistency: README lists a scripts/ folder with Python scripts, but those script files are not present in the provided file manifest; that could be an omission (docs-first intent) but the discrepancy is worth confirming.
✓ 指令范围
SKILL.md and reference docs keep the scope to read-only analysis, require explicit approval before any write action, and instruct agents to limit reads to necessary data. There are no instructions to read unrelated system files or exfiltrate data. Note: the 'browser fallback' path involves opening Meta Ads Manager in a browser context — that can implicitly access browser sessions/cookies, so confirm the approved browser context is isolated and does not leak other account data.
✓ 安装机制
Instruction-only skill with no install spec and no code files to execute — lowest install risk. The README's mention of scripts suggests future automation, but since those scripts are not included there's nothing being installed now.
ℹ 凭证需求
The skill does not declare required env vars in the registry metadata (primary credential: none), yet the docs explicitly reference typical credentials (META_ADS_API_KEY, account ID, ads_read/ads_management scopes). This is reasonable for a shared docs-only skill, but users must ensure tokens are provided to the agent via the environment or credential injection mechanism at runtime. Verify the agent will only request the minimal scopes needed (ads_read for reporting) and will prompt for approval before using ads_management scopes.
✓ 持久化与权限
always is false and there is no install-driven persistence. The skill does not request system-wide configuration changes or permanent presence. Autonomous invocation is allowed (platform default) but the skill's instructions emphasize read-first and explicit approval for writes.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.0.02026/3/15
首个面向 OpenClaw 代理的共享文档优先 Meta Ads 技能，内置账户结构、审计流程、优化指引与“先审批后写入”防护机制。
● 无害
安装命令
点击复制官方npx clawhub@latest install openclaw-meta-ads
镜像加速npx clawhub@latest install openclaw-meta-ads --registry https://cn.longxiaskill.com
技能文档
--- name: openclaw-meta-ads description: OpenClaw agents 共享的 Meta Ads 技能。以生产安全、先读后写的流程，分析 Meta（Facebook 与 Instagram）广告账户、推广系列、广告组、广告、成效洞察及表单线索数据。当 agent 需要 Meta Ads 报表、账户审计、创意或受众诊断、表单审核，或在正式修改广告前获取建议时使用。任何写入操作须先获得明确批准。
# OpenClaw Meta Ads 跨 agent 处理 Meta Ads 时使用本技能。此为共享技能，非 Poseidon 专属。账户级 token、账户 ID 与业务规则须置于技能核心之外。
使用场景
- 账户与推广系列效果报表
- 成效洞察分析
- 推广系列、广告组、广告诊断
- 表单与线索流程审查
- 创意与受众审查
- 正式改动前的结构化建议
默认立场
先只读。先分析，再建议，最后才改动。未经明确批准，不得对线上 Meta Ads 做任何修改。访问模型
Meta 账户访问按 agent 区分。技能共享，但各 agent 需使用自己的 token、账户 ID 与权限。如凭证或权限缺失，立即停止并修复访问，切勿假装可执行真实操作。设置与参考
仅读取所需内容：references/api-setup.md— token、权限与连接基础references/account-structure.md— 推广系列、广告组、广告、创意、洞察、线索结构references/insights-queries.md— 常用查询模式references/audit-workflows.md— 实操审计流程references/optimization.md— 优化启发与常见错误references/browser-fallback.md— 仅当 API 不可用或需 UI 确认时
安全规则
- 绝不在文件或对话中暴露 token 或密钥
- 优先只读分析，再做运营变更
- 暂停、启用、编辑、预算或创意修改须获明确批准
- 将优化建议视为启发，而非绝对真理
- 必要时先清除线索数据中的 PII，再进行推理
输出风格
先给发现与建议行动。报告保持实用：- 正在发生什么
- 哪里看起来有问题
- 下一步该查什么
- 如获批准，该改什么