DataDoe AI Skills
Public library of reusable AI skills that help users go from one prompt to a working MVP/PoC app powered by the DataDoe API.
Each skill is a practical blueprint you can use in coding agents (for example Cursor, Codex, Claude Code, Aider) to scaffold a ready-to-run application flow.
Most skills run in any MCP client (Claude, ChatGPT, Cursor, Claude Code) against the DataDoe MCP server; a few scaffold full apps via the DataDoe REST API. The Interface column below tells you which each skill needs.
Purpose
- Provide ready-made skill instructions for building DataDoe-powered apps quickly.
- Standardize how AI agents generate apps that integrate with the DataDoe API.
- Reduce time from idea to first working prototype.
How To Use
- Pick a skill from the
skills/ directory.
- Open its
SKILL.md file.
- In your coding agent, ask it to build the app using that skill.
- Follow the generated setup steps (environment variables, API key, run commands).
- Run and iterate on the generated MVP/PoC.
One-shot example prompt
Build this app using @skills/create-orders-manager/SKILL.md.
Follow the skill exactly and generate a complete working implementation.
Available Skills
Access: 🔍 Read-only - only reads your data, never changes anything on Amazon. ✍️ Write (approval-gated) - can change your account, but always previews with a dry run first and applies only on your explicit approval; you stay in full control.
Interface: 🔌 MCP - runs in any MCP client (Claude, ChatGPT, Cursor, Claude Code). 🧩 API - builds on the DataDoe REST API. 🔀 Both - MCP-first with a REST fallback.
Output: 📊 Report - analysis / an HTML card / a dashboard, delivered in your assistant. 🖥️ App - a runnable code project the skill scaffolds. ⚡ Action - an approval-gated change to your account.
| Skill | Access | Interface | Output | Category | What it builds | Best for |
|---|
create-orders-manager | 🔍 Read-only | 🧩 API | 🖥️ App | Reporting | A complete single-page Orders Manager app with seller/vendor picker, orders table with nested line items, filtering, local tagging, and robust DataDoe export/poll/download flow. | Teams that need a fast operational dashboard for order analysis and workflows. |
create-amazon-reconciliation-dashboard | 🔍 Read-only | 🔀 Both | 📊 Report | Reporting | A self-contained interactive Amazon Reconciliation Dashboard with month switching, KPI cards, charts, a daily summary table, and an Order Explorer for reconciling Amazon orders against settlements. | Teams that need to compare Amazon orders and settlements across a 6-month window with clear cross-month reconciliation. |
weekly-sales-briefing | 🔍 Read-only | 🔌 MCP | 📊 Report | Reporting | A polished interactive HTML weekly sales briefing card with KPI tiles, top 5 SKUs, biggest week-over-week drops, and concise AI-generated insights from DataDoe exports. | Sellers and operators who need a fast weekly performance recap with clear trend signals and SKU-level highlights. |
sales-movers-scanner | 🔍 Read-only | 🔌 MCP | 📊 Report | Reporting | A catalog-wide scan of the biggest sales movers (up and down) that decomposes each move into its cause - traffic, conversion, price or buy-box - ranked by money, with a data-completeness guard so a lagging recent week isn't reported as a fake collapse, from DataDoe sales & traffic data. | Sellers who want a Monday-morning "what changed and why" list that's already diagnosed per SKU, not just a table of deltas. |
vendor-performance-dashboard | 🔍 Read-only | 🔌 MCP | 📊 Report | Reporting | A live Vendor Central performance dashboard - KPI scorecards for vendor sales, ad spend, ad sales, ROAS and orders (last 30 days vs the previous 30, anchored on the newest vendor data), a daily sales and ad-spend trend, and an ASIN table with sales, ordered units, glance views and CVR - as a Cowork live artifact or an HTML file, from DataDoe. Offers an updated version with ordered revenue, TACoS, and lost Buy Box and out-of-stock rates per ASIN. | Vendors and agencies who want a quick, shareable view of the main 1P performance factors with one-click refresh. |
Action skills use actions_start for Amazon lookups, write dry runs, and approved changes. All MCP clients use the standard /mcp/v1 endpoint. Use actions_details_schema_get to check whether an Action reads or writes Amazon data; its access field reports READ or WRITE. A read Action, including Amazon Ads FIND, runs on a Seller or Vendor with Read only or Read and write access. A live write Action needs Read and write. A dry run still validates a write Action on a Read only account and does not queue it.
Installation with skills CLI
To install selected skill from this repository use the following command:
npx skills@latest add Deltologic/datadoe-ai-skills --skill SKILL_NAME
External Links
Contact
For questions, feedback, or collaboration requests, email: [email protected]
