Skip to content

AI agent skill

Terminal window
npx skills add beexar-games/public-api

That installs the beexar-integration skill into every agent on your machine that supports skills — Claude Code, Cursor, Codex, OpenCode, Gemini CLI, GitHub Copilot and around eighty others. Then just ask:

Integrate Beexar into this project.


  1. Reads your codebase first. Language and runtime pick the SDK; the project type decides whether you need the wallet callbacks, the launcher, or both; and it looks for the ledger the callbacks will have to run inside.

  2. Reads only what it needs. The skill ships a reference per language plus focused notes on the wallet callbacks, error codes, timings and the launcher.

  3. Implements against your code. Credentials from environment variables, the four callbacks wired into your existing router, one database transaction per request.

  4. Tells you what to verify. Which files changed, which environment variables to set, and to run the Integration Test Game — the only real proof the integration is correct.

Most integration skills are a snapshot of a document someone wrote once. This one is generated from the platform’s own source on every release:

Part of the skillGenerated from
the api_code registrythe platform’s own code table
the timing and retry budgetsthe gateway’s retry configuration
the contract versions it quotesinfo.version in the four OpenAPI documents

Our CI fails if any of them drifts. And the skill itself carries the contract versions it was built against, with instructions to check them against the live specs and to tell you — rather than invent a method name — when they disagree.

  • It will not implement idempotency inside the SDK. That has to live in your database transaction, and the skill says so instead of pretending otherwise.
  • It will not turn off signature verification to make something work.
  • It will not use floating-point arithmetic for money.

Every agent can still read the docs directly:

https://docs.beexar.com/llms.txt — a map of this site for language models, and /llms-full.txt with the full text of every page in one file.