Agent access. Live now

Your AI agent can run SKUASCOUT

SKUASCOUT ships a remote MCP server. Connect Claude Code, Cursor, or any MCP-capable agent, and it scans niches, runs analyses, and reads your reports on your account: your identity, your credits, the same tools you use in the app.

Included on every paid planWorks with Claude Code and CursorSame tools as the appYour keys, your credits

Start free See pricing

What delegation looks like

YouScan for pet niches under 500 reviews.
Your agentScan started on your SKUASCOUT account. It takes a few minutes; I will poll it and report back.
Your agentDone: 214 candidates, 3 look worth a viability run. Start the first?

How it plugs in

Bring your own agent

One endpoint, an open standard

MCP is the open standard that lets AI agents call tools. SKUASCOUT ships a remote MCP server, so Claude Code, Cursor, or any MCP-capable agent can work your account the way you would. No SDK and no glue code: an endpoint and a key.

Personal access keys, shown once, revocable anytime
Read-only keys for agents that should never spend
Same tools, your identity, your credits

What it can do for you

Tell your agent what you want in plain language and it maps the request onto the tools: list your scans and niches, read a viability, profitability, or deep dive report, start a scan or an analysis, look up a tariff code, cancel a running job. Every tool description carries its live credit cost, so the agent can tell you what a run will cost before it starts one.

Reads your scans, niches, and reports
Starts scans and analyses, cancels running jobs
Sees the live cost of every action before it runs it

Built with guardrails

Delegation without the anxiety

Charge before run, refund on failure
Every action is paid up front from your balance and refunded if it fails or gets canceled. An agent can never spend past what you hold.
One heavy job at a time
Agents share the same job slot as the app. No parallel fan-out burning credits while you sleep.
Keys you control
Scoped, hashed at rest, shown once. Read-only keys cannot spend a single credit. Revoking a key cuts its agent off immediately.
Everything on the record
Every tool call lands in your account's audit trail: which tool, which key, what outcome. Rate limits apply per key and per account.
Billing stays human
No agent can upgrade your plan, buy credits, or touch payment settings. Money decisions remain yours, in the app.

For developers: the endpoint and the tools

Setup in three steps, the access keys, and the full tool surface.

The tool surface

What your agent can do

A small, curated set of tools, not a raw API dump. Reads are free. Actions cost exactly what they cost in the app, charged before the run and refunded if it fails or you cancel.

Read. Free

list_searchesYour recent scans, newest first
list_nichesFilter by keyword or analyzed only, with real totals
get_nicheOne niche with its stage statuses and metrics
get_reportMarket viability, profitability, or deep dive
get_job_statusWhat is running on the account right now
get_credits_balanceBalance, plan, and the full price list

Run. Costs credits

start_niche_scanFinds candidate niches from your criteria
run_viabilityScores a niche's page-1 market against your thresholds
run_profitabilityBuilds a full Amazon P&L, with or without supplier sourcing
run_deep_diveTears down the competitor listings you pick
lookup_hs_codeClassifies a product tariff code and duty rate, or reads the rate for a code you already have
cancel_active_jobStops the running job and refunds it

Setup

Connected in three steps

Have a paid plan

Agent access is included on Platinum, Diamond, and Scale. No add-on, no separate tier.

Create a key

In the app: Settings, then Agent Access. Name the key, pick read-and-run or read-only, copy it once. You can hold up to five and revoke any of them instantly.

Point your client at the endpoint

For Claude Code it is one command. Any other MCP client needs just the endpoint and your key as a bearer token.

Endpoint

https://veezicmwktpomerosiif.supabase.co/functions/v1/mcp

Claude Code

claude mcp add --transport http skuascout \
  https://veezicmwktpomerosiif.supabase.co/functions/v1/mcp \
  --header "Authorization: Bearer YOUR_KEY"

Cursor and other MCP clients: same endpoint, header Authorization: Bearer YOUR_KEY.

Questions

The short answers

Does agent access cost extra?

No. It is included on every paid plan. Actions cost the same credits as running them yourself, charged before the run and refunded if it fails or you cancel. Your agent can read the live price list at any time, so it can tell you what a run will cost before it starts one.

Which clients work today?

Claude Code, Cursor, and any MCP client that can send a bearer token over HTTP. Browser-based connectors such as claude.ai on the web are on the roadmap.

How do long analyses work?

Scans and analyses run in the background. Your agent starts one, gets a job id, and polls status until it finishes, the same lifecycle the app uses.

Can a key see anything beyond my account?

No. A key carries exactly your identity and your permissions. Ownership is enforced on every read and every action, server-side.

Give your agent a research platform

Start free, look around, and when you are ready to put an agent to work, any paid plan unlocks the full surface.

Start freeIncluded on Platinum, Diamond, and Scale.