Skip to content

Support Boundary

Use this page when you need the shortest honest answer about support.

It answers three team questions:

  • what is safe to recommend by default
  • what is supported, but should be chosen on purpose
  • what is still experimental and should not quietly become team policy

Safe Defaults

These are the safest defaults today:

  • Go is the recommended default runtime path.
  • validate --strict is the main readiness gate for local Python and Node runtime repos.
  • Codex runtime Go, Codex package, Gemini packaging, Gemini Go runtime, and the Claude default stable lane are the main Recommended production lanes.
  • Python and Node are supported non-Go paths and the recommended non-Go choice when the local interpreted runtime tradeoff is intentional.

How This Maps To The Formal Contract

The public docs use three simple words first:

  • Recommended usually maps to the strongest current public-stable production lanes.
  • Advanced means a supported surface with a narrower, more specialized, or more careful contract.
  • Experimental means opt-in churn outside the normal compatibility expectation.

When a team needs exact policy language, the formal terms win: public-stable, public-beta, and public-experimental.

If you need the practical answer, start here:

  • Claude is recommended on the default stable hook path.
  • Codex is recommended both for the Notify runtime path and the official codex-package path.
  • Gemini packaging is recommended, and the promoted Gemini Go runtime is also production-ready.
  • OpenCode and Cursor are repo-owned integration setup paths. They are useful, but they are not the default executable runtime start.

Advanced Surfaces

Choose advanced surfaces only when the tradeoff is explicit and worth it.

Typical examples:

  • OpenCode and Cursor when the repo should own integration config instead of shipping a runtime path
  • narrower or specialized runtime expansions beyond the main recommended paths
  • install wrappers when the real concern is CLI delivery, not runtime APIs or SDKs
  • specialized configuration surfaces that are useful, but not the first default for most teams

Experimental Surfaces

Treat experimental areas as opt-in and high-churn.

They can be useful for early adopters, but they should not quietly become a long-term standard for the team.

Practical Rule

If you are choosing for a team, standardize the narrowest path whose promise you are actually willing to defend in CI, rollout, and handoff.

Public docs for plugin authors and integrators.