Structured Listing Engineering · Workshop Takeaway
The Helix + Claude Prompt Library
Five ready-to-use templates. Copy one, paste it into Claude with your Helix connection active, and adjust the brackets for your own brand.
Five ready-to-use templates. Copy one, paste it into Claude with your Helix connection active, and adjust the brackets for your own brand.
HELIX IMAGE GENERATION CONTROLLER, FINAL Brand: [BRAND NAME] Product: [PRODUCT NAME] ASIN: [ASIN] Marketplace: [US / UK] Report Type: ImageAnalysis Report Section: Carousel Insights & Recommendations PREREQUISITE This workflow requires the assistant to have visual access to the current live Amazon carousel for the specified ASIN, either through browsing access or manually uploaded reference images. If neither is available, Style Reference Mode B or C cannot be used; proceed with Mode A only, and expect looser visual consistency. OBJECTIVE Generate a premium Amazon image carousel using the AI Image Prompt from the Helix ImageAnalysis report as the authoritative image generation specification. IMAGE GENERATION SETUP Before beginning, ask me to confirm: Brand: [BRAND NAME] / Product: [PRODUCT NAME] / ASIN: [ASIN] Style Reference Mode: A. Brand Strategy, use the completed Helix Brand Strategy as the creative guideline. This source is narrative, not a visual style guide; expect softer visual consistency than Mode B or C. B. Current Listing Baseline, use the brand's own existing product images from the current live listing, not competitor images that also appear in the same report section. Confirm you are referencing the brand's own photos only before proceeding. C. Hybrid, Recommended. Brand Strategy for brand DNA, Current Listing Baseline for product-specific visual execution. Image Source: AI Image Prompt only Workflow: One image at a time AUTHORITATIVE IMAGE SPECIFICATION For each carousel image, extract ONLY the AI Image Prompt. Treat it as the authoritative generation specification. Ignore all other report fields (Image Title, Headline Copy, Supporting Copy, Priority Message, Research Insight, Visual Description, Technical Requirements). Do not merge, rewrite, reinterpret, or expand from these fields. TEXT OVERLAYS If the AI Image Prompt explicitly specifies text overlays, include only those exact overlays. Do not rewrite, improve, shorten, expand, or invent copy. PREVIEW STEP Before generating any image, display the extracted AI Image Prompt exactly as stored. Do not summarize, rewrite, or optimize. Wait for my response: Approve / Edit Prompt / Skip. GENERATION STEP When I reply "Approve": generate only the current image, one image only, then stop immediately. CONTINUATION After each image, stop and wait. If I say Continue, Next, or Run Image [N]: reuse the same report and Style Reference Mode, retrieve the next AI Image Prompt, display it, wait for approval, generate one image, stop again. IMAGE RULES Generate one square Amazon carousel image at a time. Never generate multiple images, collages, or contact sheets. Never invent marketing claims, badges, certifications, icons, statistics, review stars, props, or lifestyle elements unless explicitly instructed inside the AI Image Prompt. Preserve product packaging, flavor, branding, and product identity. PRIORITY ORDER 1. AI Image Prompt 2. Selected Style Reference Mode 3. Brand assets BEGIN Ask me to select A, B, or C. Then begin with Image 1.
HELIX BULK BULLET REFORMAT CONTROLLER OBJECTIVE Reformat bullet copy across multiple completed Helix transformations to a new character limit, while preserving structured listing engineering priority: the load-bearing claim survives, supporting detail gets cut first. SETUP Ask me to confirm: - Brand: [BRAND NAME] - Products / ASINs to include: [list] - Target character limit per bullet: [e.g. 250] TASK 1. For each product, retrieve the latest completed Helix transformation's bullets. 2. For each bullet, separate the load-bearing claim (the specific number, certification, or benefit that answers a documented Rufus/buyer question) from supporting detail (usage tips, secondary occasions, caveats, restated context). 3. Compress each bullet to the target character limit, keeping the load-bearing claim first and cutting supporting detail only. 4. Apply the identical compression method to every bullet, across every product. Do not vary the approach product to product. 5. Do not alter titles unless explicitly requested. 6. Present results as a Before/After table, grouped by product, for review before anything is finalized. 7. Default deliverable: a CSV file, one row per product, columns exactly: ASIN | PRODUCT | BULLET 1 | CHARACTER COUNT | BULLET 2 | CHARACTER COUNT | BULLET 3 | CHARACTER COUNT | BULLET 4 | CHARACTER COUNT | BULLET 5 | CHARACTER COUNT RULES - Never drop a certification, safety claim, or verified number that answers a documented buyer question. - Never invent a claim that wasn't in the original bullet. - Preserve as much keyword coverage as the character limit allows. - If a bullet cannot hit the limit without losing a load-bearing claim, flag it instead of silently cutting it.
HELIX TO WALMART REFORMAT CONTROLLER OBJECTIVE Take a completed Helix transformation (Amazon-optimized copy) and rebuild it to comply with Walmart Marketplace's content requirements, preserving the underlying structured product knowledge rather than rewriting from scratch. SETUP Ask me to confirm: - Brand: [BRAND NAME] - Product / ASIN: [PRODUCT] - Which completed Helix transformation to use as the source WALMART RULES TO APPLY Title: 150 character hard cap, 50-75 recommended. No promotional language, no special characters, single brand/product. Key Features: 3 to 10 features, most important first. 80 characters max per feature, spaces included. No HTML, no numbered lists, no exclusivity claims. Description: One flowing paragraph, no bullet points, no HTML tags. 150 word minimum. No promotional claims, no external URLs, no emojis. TASK 1. Retrieve the source transformation's title, bullets, and description. 2. Check the title against Walmart's cap. Rewrite only if it exceeds 150 characters; otherwise carry it forward unchanged. 3. For each Amazon bullet, extract the single load-bearing claim and compress it into a short phrase, 80 characters or fewer. Preserve priority order from the original bullet sequence. 4. Merge the description into one paragraph. Remove HTML breaks and list structure. Scan for and remove any promotional language Walmart prohibits (best-selling, premium quality, top rated, clearance, etc). 5. Present Before (Amazon) and After (Walmart) side by side for review before finalizing anything. RULES - Never drop a certification, safety claim, or verified number. - Never invent a claim that wasn't in the original copy. - Flag anything that cannot fit the character limit without losing a load-bearing claim, rather than silently cutting it.
"Pull my Amazon sales and traffic data for the last 30 days across
my catalog. Show me which listings have high sessions but a
conversion rate below my account average. Rank them by lost revenue
opportunity, sessions times the conversion gap."HELIX BRAND STRATEGY CONTROLLER OBJECTIVE Build a brand strategy document using Helix's brand and listing data as the foundation, then pressure-test every differentiation claim against real competitor listings and real customer review data before finalizing anything. SETUP Ask me to confirm: - Brand: [BRAND NAME] - Primary product / ASIN to use as reference: [PRODUCT] PHASE 1: DRAFT FROM HELIX DATA 1. Retrieve the brand record: brand story, audience positioning, voice and messaging guidance, proof points, and topics to avoid. 2. Retrieve the latest completed transformation's ListingAnalysis report: ICP and USP mapping, keyword coverage by search volume, brand voice alignment examples, and compliance guidance. 3. Retrieve the ImageAnalysis report: executive summary, what customers want, and the named competitors in its competitive landscape data. 4. The brand record, the ListingAnalysis report, and the ImageAnalysis report each generate their own independent ICP narrative. Do not present all three. Reconcile them into one unified profile. 5. Draft: Brand Overview, Positioning Statement, ICP, USP, What Customers Want, Brand Voice (with before/after copy examples pulled from the report, not invented), Proof Points, Top Keywords by search volume, and Brand Guardrails. 6. Present this draft for review before moving to Phase 2. PHASE 2: PRESSURE-TEST WITH REAL MARKET EVIDENCE 7. Cross-check every "topics to avoid" item in the brand record against the actual live transformed copy. Flag any contradiction directly, including legal or compliance risk language that's currently in use. 8. Identify named competitors from the ImageAnalysis report, then independently search Amazon's live catalog for each. Do not rely only on the pre-listed names, a live search may surface more relevant or higher-ranked competitors. 9. Pull real customer review sentiment for this product and for at least the top 2 to 3 competitors identified. 10. For every core USP claim, classify it as one of: - PARITY: competitors make the identical claim, often cheaper - VULNERABILITY: a competitor makes the same type of claim more confidently or with less risk attached - TRUE DIFFERENTIATOR: no competitor in the set has anything comparable 11. Rebuild the differentiation section around a Need, Deliver, Differentiate framework. Cite the specific competitor name, price, and category rank behind every verdict. RULES - Never assert a differentiation claim without checking it against at least one real, live competitor listing. - Never treat Helix's internal brand-consistency check as equivalent to competitive market validation. They answer different questions. - Flag contradictions between brand guardrails and live copy plainly. Do not soften or bury a real compliance risk. - Cite the specific source, a review quote, a competitor ASIN, a price, a rank, behind every finding. No unsourced claims.
One connection, five templates, and your own brand's data. Adjust the brackets and go.