Is the Check E-Bike Return Risk skill safe to inspect before installing?

Yes. Every packaged file is public, the archive inventory and SHA-256 digest are published, and the caller makes only a visible HTTPS GET request when an estimate is requested. URL-only mode makes no request.

Is the e-bike return-risk agent skill free?

Yes. The skill, source-linked estimate, policy dates, and seller citations are free. Skinflint's separate buyer kit is optional.

Free agent skill · maintained policy data

Give a shopping agent an exit calculator, not another opinion.

The Check E-Bike Return Risk skill lets an AI agent calculate ordinary return eligibility, known deductions, and estimated refunds from Skinflint's source-dated policy records. It preserves unpriced freight as unknown and cites the seller's first-party policy.

This is not a prompt wrapper. The useful layer is a maintained, deterministic data service: 13 direct-order U.S. e-bike policies, normalized arithmetic, source dates, and failure rules that stop an agent from quietly turning “unknown” into “free.”

What the skill answers

Can this direct-order bike still be returned after a test ride?
What published restocking fees and fixed charges apply at the actual price?
What refund remains before any unpriced freight or inspection adjustment?
Which return window, ride limit, packaging rule, and first-party source support the answer?

Install the portable skill

Download one inspectable ZIP containing the complete skill folder. No installer runs, no credentials are requested, and no shopper data is bundled.

Download the complete skill Read SKILL.md first Inspect manifest

Package files: deterministic caller · API contract and supported brands · agent interface metadata

Where to put it

ChatGPT

On an eligible workspace, open Plugins → Skills → Create → Upload from your computer, then select the ZIP. ChatGPT scans an uploaded skill before making it available. OpenAI's current skill-upload guide has the account and workspace details.

GitHub Copilot in VS Code

Extract the folder into .github/skills/check-ebike-return-risk/, .claude/skills/check-ebike-return-risk/, or .agents/skills/check-ebike-return-risk/ inside the repository.

OpenCode or another open-standard client

Extract it into the client's project or personal skills directory—for example, .opencode/skill/check-ebike-return-risk/. The directory that directly contains SKILL.md must stay named check-ebike-return-risk.

Review SKILL.md, the caller, and the API contract.
Place the complete folder where the agent discovers skills.
Ask the agent to use $check-ebike-return-risk with a supported brand, USD price, and either unopened or ridden.

Try a request without calling the API

node scripts/estimate.mjs aventon 1500 ridden --url-only

The full machine contract is also available as OpenAPI 3.1. Runtime requests require no account or API key. The agent should always lead with eligibility, preserve unpriced freight, include the checked date and source, and tell the buyer to confirm the live policy at checkout.

Supported launch records

Aventon, Lectric, Velotric, Rad Power Bikes, Ride1Up Revv 1, Fiido, Himiway, HOVSCO, and Ecotric. Records cover the scope stated by the API, generally direct U.S. orders—not dealers, marketplaces, used sales, warranties, or other countries.

Example answer

A ridden $1,500 Aventon has an $875 known return deduction and an estimated $625 refund under the indexed ordinary-return policy. That combines the published 50% used-bike deduction and $125 label charge. The agent still reports scope, condition, packaging, checked date, and first-party source rather than treating one number as the whole return decision.

Need a human-readable source receipt?

Every result links to a free brand breakdown. Buyers who want an editable seller-specific cancellation or return request, private refund downside file, route-specific evidence checklist, applicable calendar deadline, and one-brand 30-day source watch can also use the $12 Return Rescue Kit. Payment never changes the public estimate or source ordering.

Is the skill safe to inspect before installing?

Yes. Every packaged file is linked above, the archive inventory and SHA-256 digest are published in the manifest, and the caller makes only a source-visible HTTPS GET request when an estimate is requested. Use --url-only to inspect the encoded request without contacting Skinflint.

Is the skill free?

Yes. The skill, source-linked estimate, policy dates, and seller citations are free. The separate $12 buyer kit is optional and never changes the public arithmetic.