Amazon Skills: 53 Markdown Instruction Files for Seller Agents
Free AI agent skills for Amazon sellers— keyword research, competitor analysis, listing audit & more. Works with OpenClaw, Claude Code, Cursor, Windsurf, Codex and any agent that supports the Skills format.
At a glance
- What is it?
- Nexscope ships Amazon seller expertise as plain-text skill files that any Skills-compatible agent can read. The install path is one npx command, but the skills carry no data pipeline, no API keys, and no live marketplace feed, which shapes both what they are good for and where they stop.
- Who is it for?
- Adopt Amazon Skills if you already run a Skills-compatible agent and want Amazon-specific reasoning frameworks loaded without writing prompts yourself, and if you are comfortable supplying the marketplace data the skills reason over. Do not adopt it if you need scheduled competitor tracking that fires on its own, or if you expect the FBA and tariff calculators to pull live fee schedules, since the README describes instruction files rather than data services.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 21 days ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
What the Amazon Skills repository actually contains
This is a collection of 53 instruction files written for AI agents, aimed at people who sell on Amazon and already use an agent that understands the Skills format. The README describes them as "plain-text instruction files that give AI agents specialized expertise in selling on Amazon," with no binaries and no API keys. The repository is Python-tagged, MIT-licensed, and the README lists no releases, so installation is from the main branch rather than a versioned tag. The collection is grouped by job: product research and keywords, competitor analysis, pricing and profitability, advertising, and further categories the truncated README does not fully show. Twelve marketplaces are listed: US, UK, DE, FR, IT, ES, JP, CA, AU, IN, MX and BR. The gap this fills is not data access. It is the reasoning layer. An agent asked to score a keyword or build a PPC structure has no built-in model of how Amazon selling works, and these files supply that model as text the model can read at inference time.
How a skill file reaches the model
The mechanism is deliberately thin. Each skill lives in its own directory named after the skill, for example amazon-keyword-research or amazon-fba-calculator, and the README's install examples reference those names directly. There is no server, no plugin runtime and no tool registration described. The agent reads the markdown, and the markdown tells it how to approach the task. That means the quality ceiling is set by the instructions themselves plus whatever data the user or the agent can reach. The README's keyword example asks for "search volume, competition, and opportunity score" for a term on Amazon US, which implies the agent either has a data source or the skill explains how to reason from what the user pastes in. The README does not say which. Treat the skills as analytical scaffolding, not as a query engine, until you read the individual skill directories and confirm what inputs they expect.
Installing one skill versus all 53
Installation runs through the skills CLI. The README gives a global install of the whole collection: npx skills add nexscope-ai/Amazon-Skills -g. A single skill uses the --skill flag with the directory name, for example npx skills add nexscope-ai/Amazon-Skills --skill amazon-keyword-research -g. The -g flag places skills globally for the agent rather than per project. Because there are 53 of them, the single-skill path matters: loading every Amazon skill into an agent's context for a session that only needs FBA math adds instructions the model has to sift through. The README does not document an update or uninstall command, and there are no releases to pin to, so re-running the add command against main is the only upgrade path visible in the material. If you need reproducibility across a team, that is a real constraint worth testing before you standardise on it.
Where the collection is strongest and where it is thin
The README marks skills with two states: Available, described as production-ready, and Beta, described as functional and being improved. Every skill visible in the truncated list carries the Available mark, which either means the beta set sits in categories the excerpt cuts off or that the status column is not currently doing much work. The most concrete skills are the ones with defined arithmetic: amazon-fba-calculator for referral fees, fulfillment fees, storage and net margin; amazon-shipping-calculator for dimensional weight and removal costs; tariff-calculator-amazon for duties, landed cost and VAT/GST. Fee and duty tables change, and a text file cannot update itself. The README gives no versioning, no effective date and no maintenance note for those numbers. That is the single biggest thing to verify by opening the files. The research-oriented skills, such as amazon-niche-finder and amazon-sales-estimator, depend more on the model's reasoning and on the data you feed it, so their output quality will track your inputs more than the file's age.
Amazon Skills against a seller analytics platform
The obvious comparison is a hosted Amazon analytics product, the kind that connects to Seller Central or scrapes the marketplace on a schedule and stores history you can chart. Those tools own the data pipeline: they collect, store and refresh. Amazon Skills owns none of that. It is instructions. The practical difference shows up in two places. First, monitoring: amazon-competitor-monitoring is described as real-time competitor tracking covering pricing changes, inventory alerts and review velocity, but a markdown file cannot poll anything on its own. The tracking happens only when you invoke the agent and only against data the agent can reach. Second, history: a platform can show a price chart over ninety days; a skill can only reason about the snapshots you give it. The trade is deliberate. You give up collection and storage, and in return you get no subscription lock-in, no marketplace credentials handed to a third party, and the ability to edit the instructions yourself under MIT. For sellers who already have data exports, that is a reasonable swap. For sellers who want alerts without asking, it is not.
Licence, maintenance and the cost of keeping skills current
The repository is MIT-licensed, which permits commercial use, modification and redistribution provided the copyright notice and permission notice are retained. That matters here because the skills are files you can fork and edit. If your fee assumptions differ from the shipped ones, you can change them locally. The flip side is that forking is also the maintenance model: there is no release cadence in the material, only a main branch last pushed in August 2026. Every fee table, every marketplace rule and every Amazon policy reference inside these files is a candidate for drift, and nothing in the repository signals when a file was last checked against Amazon's published rates. The README points to a companion repository, eCommerce-Skills, with 142 skills spanning Amazon, Shopify, eBay, Etsy, TikTok Shop and Walmart, and to a Nexscope skill hub, so the project is part of a larger catalogue rather than a standalone tool. This is not legal advice; if you redistribute modified skills inside a product, read the MIT text and your own obligations rather than relying on this summary.
Who should install this and what to check first
Install it if you run OpenClaw, Claude Code, Cursor, Windsurf, Codex or another Skills-compatible agent and you want Amazon-specific analysis without authoring the prompts. The keyword research, listing optimization and PPC skills are the ones the README leads with, and they are the ones where a structured framework saves the most time. Skip it if your workflow depends on unattended monitoring, on historical price or BSR series, or on fee calculations that must match Amazon's current schedule to the cent without you checking them. Before you rely on any number, read the skill directory for the calculator you care about and confirm which marketplace edition it targets and what date its figures carry. Start with one skill, not 53: npx skills add nexscope-ai/Amazon-Skills --skill amazon-fba-calculator -g, then run it against a product whose true margin you already know. If the output matches your own math, the rest of the collection is worth loading. If it does not, you have found the drift before it reached a pricing decision.
Editorial conclusion
Adopt Amazon Skills if you already run a Skills-compatible agent and want Amazon-specific reasoning frameworks loaded without writing prompts yourself, and if you are comfortable supplying the marketplace data the skills reason over. Do not adopt it if you need scheduled competitor tracking that fires on its own, or if you expect the FBA and tariff calculators to pull live fee schedules, since the README describes instruction files rather than data services. Before committing, open the amazon-fba-calculator and tariff-calculator-amazon skill directories and check whether the fee tables are dated and which marketplace edition each one targets, because that is the part most likely to drift from Amazon's published rates.
Community notes