{"id":71221,"plugin_id":"plugin_asdk_app_6ad0f3b6651881918d0a5823d51dd6eb","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-10-29T19:13:34.603Z","digest":"1c7dfaf313e6afed43702d6b1336068df849068715f83241869741eff1f0e3bb","against":null,"payload":{"description":"Use when the user asks about their AgentFabrica marketing panel, such as brands, the content calendar, items waiting for approval, ad pre-test results, lead or WhatsApp counts, plan quota, or wants a draft post added to a brand's approval queue.","included_files":[],"name":"agentfabrica-panel","skill_md_contents":"---\nname: agentfabrica-panel\ndescription: Use when the user asks about their AgentFabrica marketing panel, such as brands, the content calendar, items waiting for approval, ad pre-test results, lead or WhatsApp counts, plan quota, or wants a draft post added to a brand's approval queue.\n---\n\n# Working with the AgentFabrica panel\n\nAgentFabrica is a marketing operations panel. The connected tools read the user's own account and can add a draft post to an approval queue. They never publish, approve or delete anything.\n\n## Start with the brand\n\nMost tools take a `brandId`. When the user names a brand, or the account may hold more than one, call `list_brands` first and use the id of the matching brand. If two brands could match the name, ask which one the user means before going further.\n\n## Pick the tool that answers the question\n\n| The user wants | Tool |\n|---|---|\n| Their brands, industry, website and tone of voice | `list_brands` |\n| Upcoming or past posts and their status | `scheduled_posts` |\n| Everything waiting for a person to decide | `pending_approvals` |\n| How ads scored with the panel before launch | `ad_pretests` |\n| How reliable the panel has been for this brand | `calibration_report` |\n| Which hooks, angles and formats have worked for the brand | `learning_pool` |\n| Lead counts by status and source | `lead_summary` |\n| WhatsApp line status and conversation counts | `whatsapp_summary` |\n| Plan usage and what remains | `quota_status` |\n| Recently generated content | `generated_assets` |\n| A new draft post in the approval queue | `submit_post_for_approval` |\n\n## Report results accurately\n\n- Approval items come with a `reviewUrl`. Give the user that link, and say that approving or rejecting happens in the panel.\n- `lead_summary` and `whatsapp_summary` return counts only. If the user asks for names, phone numbers or message text, explain that the panel does not share them through this connection.\n- An ad pre-test score is the panel's reaction to the ad. Do not present it as a forecast of reach or sales. When the ad later ran, the real result is included, so report both.\n- When `calibration_report` or `learning_pool` says there is not enough data, pass that on instead of estimating a number.\n- If a tool returns an error, tell the user what failed and what they can do next. Do not fill the gap with guessed figures.\n\n## Adding a draft post\n\n1. Confirm the brand, the platforms (facebook, instagram, linkedin) and the caption with the user before calling `submit_post_for_approval`.\n2. Instagram needs at least one public `https://` image URL. If the user has none, offer Facebook or LinkedIn, or ask for an image link.\n3. Without `scheduledAt`, the brand's best posting time is used. Say so in the reply.\n4. After the call, tell the user the draft is waiting for approval and share the `reviewUrl`. Never describe the post as published or scheduled to go out on its own.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}