gooseworks-ai/goose-skills / skills/shared/launch-meta-ad-campaign
launch-meta-ad-campaign
VerifiedTake someone from a business goal to a verified, paused-then-live Meta/Facebook/Instagram campaign without making them learn Meta's hierarchy. Use automatically for requests to launch, run, create, publish, set up, or plan Meta ads; use reporting skills for analysis-only requests.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 22.2 KB |
| agents/openai.yaml | config | 255 B |
| references/direct-meta-adapter.md | markdown | 4.4 KB |
| references/gooseworks-adapter.md | markdown | 4.1 KB |
| references/planning-only.md | markdown | 2.7 KB |
| skill.meta.json | config | 961 B |
| tools/meta_marketing_api.js | script | 38.2 KB |
references/gooseworks-adapter.md
View on GitHub# GooseWorks adapter
## Summary
Use this adapter when the GooseWorks launch tools are available. It keeps the campaign, creative, Meta context, approval page, push, and recovery state linked. The approval page is the only authorization mechanism for a GooseWorks write.
## Required tool surface
Use the consolidated tool names when available:
- `campaign_read`
- `ads_creative_read`
- `meta_status_get`
- `list_meta_entities`
- `get_meta_ad_context`
- `meta_read` (view `pages` for the publishing Page, view `push_status` with `push_id` for a push)
- `prepare_meta_ad_push`
Older split read names may exist for compatibility, but do not declare them as installable shared tools. `prepare_meta_ad_push` and `meta_read` view `push_status` are the write-workflow boundary.
## From an outside host
From Claude Code, Cursor or any MCP client, `meta_read`, `prepare_meta_ad_push` and `refresh_meta_sync` take **`brand_id`** on every call: the GooseWorks brand id from `brand_read` without `brand_id`. Inside the GooseWorks coworker they take no brand, because the session is bound to one. The calls act as the signed-in user, on a brand in their own organisation.
```json
{ "tool": "prepare_meta_ad_push", "args": { "brand_id": "<brand.id>", "render_ids": ["<render id>"],
"mode": "NEW_CAMPAIGN", "page_id": "<id from meta_read view pages>", "...": "..." } }
```
The read tools in step 3 (`meta_status_get`, `list_meta_entities`, `get_meta_ad_context`) may not be listed on an outside connection yet. Without them you cannot see existing ad sets. Do not ask the user for raw Meta ids. Offer the new paused Traffic campaign (`NEW_CAMPAIGN`) and say that reusing an existing ad set needs the brand's coworker in GooseWorks.
## Workflow
1. If the conversation contains a `push_id`, call `meta_read` with `view: "push_status"` and that `push_id` first, and resume it.
2. Use `campaign_read` and a bounded `ads_creative_read` to inspect the current campaign and reusable creative. Start with at most 10 published creatives; if none exist, inspect at most five review candidates and load full detail only for the 2–3 under consideration.
3. Use `meta_status_get` and `list_meta_entities` for the connected account. Call `get_meta_ad_context` for current ads that overlap the new intent or may be reused. Trust its Page/Instagram identity and provenance; do not ask for opaque IDs already present in the dossier.
4. Recommend reuse when an existing ad set matches objective, geography, audience, schedule, and optimization. Explain overlap before proposing a separate campaign.
5. Call `meta_read` with `view: "pages"` and use one of its ids as `page_id`. Call `prepare_meta_ad_push` only after every decision and destination/policy check is complete. Pass the source campaign ID so the push remains linked.
6. Present the exact summary and returned approval URL/card. Tell the human to choose **Approve paused launch** or **Decline**. Do not execute the push from chat.
If `prepare_meta_ad_push` returns `PREPARING`, the checks are still running: wait about 30 seconds, read `push_status`, and present the plan only once it is `AWAITING_APPROVAL`. Never prepare it again meanwhile.
7. After the human returns, call `meta_read` with `view: "push_status"` and the same `push_id`, and act on `push.state`:
- `PREPARING`: the checks are still running; read again in about 30 seconds.
- `PREPARE_FAILED`: the checks refused the plan and nothing exists in Meta; relay the reason and findings.
- `AWAITING_APPROVAL`: return the same approval link; do not create a new push.
- `APPROVED` or `RUNNING`: approved; Goose is creating the paused ads (usually a minute or two). Read again shortly; do not prepare again.
- `COMPLETED`: report every ID only after paused readback.
- `PARTIAL` or `FAILED`: list created IDs, verified states, failure, and `next_action`.
- `CANCELED`: report that nothing was created.
The push status lists every object it created, so do not run `refresh_meta_sync` after a launch. Use it only when an object you know exists is missing from reads.
Activation remains a separate GooseWorks UI flow with its own typed confirmation. Never activate during this skill or during tests.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |