Skip to content

release-engineering — guides

release-engineering is the SRE/ops outer loop: deployed end-to-end validation above work-loop's inner build loop. Where core validates that code works locally — it compiles, tests pass, the diff is clean — release-engineering validates that the deployed whole works: the integrated artifact, running in a real environment, observed through real telemetry.

The pack ships two primitives: release-lead (the supervisor agent) and release-loop (the skill doctrine). Together they cover the full arc from "deploy-ready artifact" to "prod ship ratified by a human" — autonomously on ephemeral environments, and with an explicit consent gate at the only irreversible exit.

Before you start: release-engineering hard-depends on core. Install core first.

agentbundle install --pack core           # required first
agentbundle install --pack release-engineering

New here? Read The release loop explained for the why — the inner/outer split and the minimum-regret autonomy carve. Then run your first release with Your first release.


Tutorials

Learning-oriented walkthroughs.

How-to

Task-oriented recipes for when you already know what you're doing.

  • Run a release — trigger release-lead, monitor convergence rounds, review the release-readiness record, and ratify at G5.

Reference

Information-oriented — dry, complete, look-it-up-when-you-need-it.

Explanation

Understanding-oriented — the why behind the design.

  • The release loop explained — the inner/outer loop split, the minimum-regret autonomy carve, the ephemeral environment model, the convergence policy, and the feedback seam between the outer and inner loops.

Cross-cutting guides — installing the catalogue, upgrading packs, the adapter support matrix — live in ../_shared/. For the full picture of how the release loop composes with discovery and build, see the three loops as a system.