Your tribal knowledge,
finally put to work.
Skills are guided, multi-step automations that hide the gnarly parts of working with systems like Amazon Marketing Cloud, Orders, Inventory, and more. Run them from chat, code, or a CLI; the heavy lifting is already done.
Pre-built playbooks.
Verified, tested, yours to bend.
Skills are guided, multi-step automations that wrap MCP tooling into something an end user can actually run. Pick one from the library, watch it work, then fork it and make it yours.
Every skill ships with a happy path, guardrails, and a full audit log. No prompt engineering. No glue code. Just the outcome you wanted in the first place.
Audience builds, attribution queries, and campaign reports without writing a line of SQL.
Pulls underperforming keywords, drafts negative lists, queues changes for approval.
Bundles new bugs by service, scores severity, and opens war rooms for anything P0.
Compose your own multi-step skill from any MCP tool. Version it, test it, ship it.
Pick a skill.
Watch it do the work.
AMC audience build, end-to-end. Six steps. Every output reproducible from the run log. Pause to edit before any step runs.
- ✓Pause and edit any step before it runs
- ✓Every output reproducible from the run log
- ✓Fork to add your own steps or guardrails
- ✓01 - Connect AMC instanceOAuth + IAM role assumed - scoped read/write
- ✓02 - Pick a goalNew-to-brand audience for Q2 launch
- ✓03 - Resolve datasetsJoins attributed events + dsp impressions
- ✓04 - Generate SQLValidated against AMC schema - dry-run executed
- ✓05 - Policy checkPII redaction - audience min-size enforced
- ✓06 - Run + deliverAudience pushed to DSP - report emailed to @growth
q2_launch_ntb_v1 - 1.84M users, pushed to DSP seat AMZ-91402. Dry-run looked clean. Report emailed to @growth.What the agents say after working with Kuudo
"Skills are tribal knowledge made executable. The MCPs reach your data; Atlas knows the playbook; skills know how to apply the playbook to this task — same structure, same thresholds, same edge-case handling, every time. Without them, every analysis is a hallucination waiting to happen. With them, the AI delivers the way you want it to."
The shortest path from generalist agent
to dependable specialist.
Skills are reusable, file-based bundles of expertise: workflows, context, and best practices that load on demand.
Domain expertise, not generic answers
Encode workflows, definitions, and guardrails. The agent follows the playbook.
Write the rules once
No more pasting the same brief into every chat. Skills load when relevant.
Stack Skills into bigger workflows
Small Skills combine into complete operating motions.
Loaded only when needed
Get the right context without ballooning every prompt.
Start from the catalog, then bend it
Use verified Skills out of the box, or fork one.
Tribal knowledge, finally portable
Your senior operator's playbook becomes a Skill anyone can run.
Concept and pattern adapted from Anthropic's Agent Skills overview.
MCP - the wiring underneath
Every Skill is composed from MCP tools. That's how a single amc.build_audience step can talk to AWS, your DSP, and your own warehouse.
Author a custom Skill
Compose steps from any MCP tool, define inputs, add approval gates, version it, and publish to your team workspace.
Read the Skill SDK docs ->Guides
How teams are building and running Skills — from AMC audiences to fulfillment audits.
Field notes from teams running Skills through Claude, ChatGPT, Cursor, n8n, and any MCP runtime — what's working, what to watch for, what to skip.
Counting New-to-Brand Customers in AMC Without Double-Counting the Same Buyer
How an Atlas-grounded AMC query counts genuine first-time buyers, why the conversion-time table is the only safe pick for a recurring workflow, and the five caveats the IQ docs scatter across four pages.
Sponsored Ads x DSP Overlap: The 4-Way AMC Analysis
How an Atlas-grounded agent runs the improved Sponsored Ads and demand-side platform overlap query in AMC, why Sponsored Products exposure means impressions here, and the caveats that keep the result honest.
Choosing the Right AMC Attribution Data Source for Custom Models
An operator asks for a custom attribution view in AMC. An agent grounded in Amazon Agent Atlas returns a five-row data-source matrix, six decision rules, and the footnotes that would have eaten the rest of the week.
Bring your Amazon knowledge into the game.
Connect ChatGPT. Pick the AMC and Sponsored Ads Skills. Stop re-explaining.
No credit card. No sales call.