A proxy vendor teaches Claude how to call its APIs

What it does

This repo packages Oxylabs’ commercial proxy and scraping services into five machine-readable “skill” files that tell AI agents which API to call, how to authenticate, and what output to expect. It is essentially glue code: a structured mapping layer between an agent’s intent and Oxylabs’ various product tiers. The README acts as a routing index, steering tasks like “fetch a protected page” or “extract YouTube subtitles” toward the correct skill directory.

The interesting bit

The product routing table treats the agent like a junior solutions architect, explicitly matching needs—CAPTCHAs, browser automation, structured JSON—to specific Oxylabs products. Shipping it all as a single Claude Code plugin turns the repository into a distribution channel, letting developers install vendor-specific proxy logic the same way they install linter rules.

Key highlights

• Five skills cover residential proxies, web unblockers, scraper APIs, headless browsers, and video data.
• Each skill specifies geo-targeting, IP rotation, session persistence, and expected data formats.
• A decision matrix maps user intents directly to the appropriate Oxylabs product.
• Ships as a Claude Code marketplace plugin and as a generic agent skill bundle.
• Requires separate environment-variable credentials for nearly every service tier.

Caveats

• The README is purely a routing index; the actual skill instructions and prompts are hidden in linked SKILL.md files not rendered here.
• Heavy credential sprawl: you need distinct username/password pairs for residential, datacenter, web unblocker, scraper, headless browser, and video services.

Verdict

Worth a look if you are already paying for Oxylabs and want your agent to stop guessing which proxy product to use. Skip it if you are not an Oxylabs customer; these skills are thin wrappers around their paid APIs.