---

name: "zhongcao-carousel-maker" description: "Create a Xiaohongshu or REDnote carousel from a post outline, product details, photo set, or style reference. Build an ordered 3:4 image sequence with a hook cover and supporting slides, clear focal imagery, matched visual direction, and headline-safe areas for product recommendations, tutorials, food notes, OOTD, travel guides, knowledge posts, and Xiaohongshu content images for one connected post story. Optionally it reads Xiaohongshu itself — the notes already running for the topic, one page of a note's top comments, and an account's own recent notes — so Xiaohongshu research, competitor note analysis and comment analysis rest on the platform instead of on guesswork."

Zhongcao Carousel Maker

Create an ordered two-to-four-slide REDnote (Xiaohongshu) carousel from a post outline, product details, photo set, or style reference. The result is a hook cover plus supporting slides with a shared visual direction and a deliberate reading order—not a set of unrelated image variants.

Choose the route

• Build a carousel from an outline: use beatra.images.generate when the post topic, slide roles, and visual direction are sufficient.
• Compose from a photo set or references: upload up to four ordered image references and use beatra.images.transform, declaring which image preserves the subject and which images guide only style, palette, or composition.
• Refine one selected slide: use beatra.images.edit with the accepted slide at images[0] and at most two normalized local edit regions.

This Skill owns a connected multiple-image note. Use zhongcao-cover-maker for one standalone Xiaohongshu cover. Do not route a carousel request to a one-cover workflow.

Plan the ordered story

Reuse the user's topic, audience, content outline, product facts, references, and visual style. Suggest two to four slide roles: a hook cover first, followed by explanation, proof, steps, comparison, or call-to-action slides. Default to a 2K 3:4 canvas and an explicit text-safe area on every slide, then include the exact slide count, order, canvas, and headline treatment in the paid-call confirmation.

Keep one subject, palette, lighting direction, and composition language across the set. Default to safe space for later overlay rather than guaranteeing rendered Chinese titles. If short in-image text is requested, freeze the exact copy in the confirmation and inspect it only when the result is visible.

Golden path

Lock the note type and an ordered slide-by-slide outline first. Freeze the shared visual anchors — one subject, palette, lighting direction, and composition language, plus any approved reference image that keeps the series consistent — before any paid generation. Then generate the 3:4 cover and supporting slides in that approved order, reusing those same anchors on every slide. Do not add extra paid calls just to produce another style set.

Verify sequence capability, then confirm one paid request

Use only this Skill's bundled scripts/mcp_client.py for all remote Beatra operations. Do not configure or call a host Beatra Connector and do not use REST/OpenAPI as a fallback. Upload local image files with the bundled client and preserve their declared order; an upload makes bytes available but does not inspect the images.

Before offering the multi-slide paid route, call beatra.models.list to verify that a currently selectable model supports the confirmed capability, count, canvas, and output_relationship: "sequence". Keep model: "auto" unless a real availability, compatibility, control, or price decision requires a specific model. If no live route supports the requested sequence, report that result and ask the user to choose either one image now or several separate paid requests. Do not silently submit independent candidates as though they were an ordered carousel.

Once the route is available, show one final confirmation freezing the prompt, slide order and count, sequence relationship, canvas, ordered references, text treatment, model, and controls. Planning and capability checks are free, and so is everything before the optional Xiaohongshu lookup, which is priced and approved on its own; the image request is paid. After approval, create one opaque stable client_request_id without user content and submit once. A changed prompt, slides, count, canvas, references or order, relationship, model, or control is new paid work needing a fresh confirmation and a fresh ID.

Track, review, and deliver

Keep the returned task_id and poll only it through beatra.tasks.get. If it is lost, use beatra.tasks.list for candidates and validate the selected one with tasks.get. Replay the byte-equivalent payload with the same ID only when the original creation response is genuinely unknown. Slow polling, connection, update, and authorization errors never create substitute paid work.

Cancel only on the user's request. If beatra.tasks.cancel returns 409, the original task continues and must be tracked. When output is visible, review the returned slide order, consistent palette and subject, each slide's focal clarity, text-safe space, confirmed canvas, and requested details. Deliver artifact links in returned order, observed dimensions, task ID, resolved model, and returned billing.net_charged_credits; offer at most one unexecuted, focused revision plan.

Reading Xiaohongshu before you write

Optional, and paid. When the connection exposes Beatra's public social lookup, this Skill can read Xiaohongshu directly instead of working from what the user remembers: one page of notes matching a keyword, one specific note the user pastes, one page of that note's top comments, and an account's profile or recent notes. Six operations, Xiaohongshu only.

Every one of them costs 60 credits, and there is no cheap operation on this platform to fall back on. The same reads cost 6 on TikTok. They cost 6 on Douyin too — except Douyin's own keyword search, which is also 60, so do not say "ten times Douyin" without naming the read. A three-step read — the field, one note, that note's top comments — is 180 credits, and every further page is another 60. Say the number before offering anything, confirm each lookup on its own before it runs, and say plainly that this Skill's own deliverable arrives either way at no cost. Offer one read, not a plan of four.

The rule is the whitelist, not a list of exceptions: a platform with no operation on it cannot be looked up from here, and another platform's notes are never presented as Xiaohongshu's. A returned image URL is not a viewed image — state a visual finding only about an image the host can actually open. Every figure that reaches the work is labelled as looked up with the time it was read, or as supplied by the user, or as missing. Nothing is estimated, and nothing is carried in from what notes in this category usually do.

See reading Xiaohongshu for the operations, the argument routes, the confirmation wording, and how a result is reported and recovered.

References by task

• Sequence payloads, ordered references, and slide edits: workflow
• First install or expired authorization: installation and authentication
• Bundled-client commands: Bundled MCP Client diagnostics
• Registration: installation registration
• Polling and returned task fields: tasks and results
• Billing and recovery: billing, errors, and recovery
• Disconnecting: uninstall and disconnect
• Update controls and integrity: automatic updates and safety

Runtime and safe automatic updates

The bundled client silently checks at most once every 24 hours during a public command. When a newer version is available, the client installs it automatically without separate confirmation only from fixed official Beatra discovery and immutable CDN paths. It verifies the archive, manifest, and every package-owned file, and replaces only files owned by this package. If an update step fails, the current installation remains usable and the original command continues. The setting persists for this installation.

python3 scripts/mcp_client.py update --auto off
python3 scripts/mcp_client.py update --auto on
python3 scripts/mcp_client.py update --check