Resources
Contributing

Contributing

We want contributions across:

  • docs corrections
  • examples
  • tests
  • runtime improvements
  • interop bridges
  • control-plane hardening
⚠️

Some of the newer agent-framework surfaces are still evolving and parts of the docs may still be incomplete or slightly behind the code. If you notice a mismatch, please tell us or send a fix. Those reports are useful, not annoying.

What to read first

Contribution standards

Product standards

  • Keep @veridex/agents useful for general-purpose agents, not only commerce agents.
  • Keep payments, identity, and trust native but optional.
  • Prefer explicit typed surfaces over magic or hidden runtime behavior.
  • Preserve interoperability where possible instead of introducing closed abstractions.

Runtime standards

  • Keep the core runtime small and inspectable.
  • New tools, hooks, and transports should be typed and documented.
  • Changes that affect runtime semantics should include tests.
  • New enterprise or control-plane behavior should explain tenancy, persistence, and failure behavior clearly.

Documentation standards

  • Update docs when public APIs or package boundaries change.
  • If a change alters framework direction, add or update an ADR.
  • Be honest about maturity. If something is still rough, say so plainly.
  • Prefer examples that show real use, not only idealized demos.

Where to contribute

AreaGood for
docs/ADRs, internal architecture notes, package and implementation docs
documentation/Public docs site pages, navigation, onboarding, examples
packages/agentsCore runtime, tools, hooks, memory, transports, policies
packages/agents-reactReact provider and hooks
packages/agents-adaptersImports, exports, and live bridges
packages/agents-openclawOpenClaw / Pi interop
packages/agents-control-planeControl-plane service, clients, and persistence
packages/agent-sdkPayment protocols, trust, identity, and commerce primitives
packages/sdkPasskeys, sessions, wallets, and chain integrations

Pull request expectations

  • explain what changed
  • explain why it changed
  • include tests when behavior changed
  • update docs when public behavior changed
  • avoid weakening existing safety or control boundaries without an ADR

Especially useful contributions right now

  • better examples for general-purpose agents
  • migration examples from LangGraph, OpenAI Agents, and OpenClaw
  • docs clarifying package boundaries
  • more transport and interop tests
  • control-plane operational hardening and deployment feedback

If you find something broken

Please report:

  • the package
  • the version or commit
  • what you expected
  • what actually happened
  • a small repro if possible

That helps much more than a generic “it doesn’t work.”