Skip to content

Practical patterns, starters & CLI tools for loop engineering with AI coding agents. Design systems that prompt and orchestrate agents (inspired by Addy Osmani and Boris Cherny). Includes loop-audit, loop-init, loop-cost.

1.5 KB
Raw Blame

Changelog

All notable changes to @cobusgreyling/loop-init are documented here.

[Unreleased]

[1.7.0] - 2026-08-28

Added

  • thin-loop pattern: one GitHub Action, no STATE.md / skills / budget files
  • Default --tool is claude

Changed

  • Companion CTAs (--with-foundry, --with-memory, --with-fleet) print only when those flags are passed

[1.6.0] - 2026-07-29

Added

  • --model-provider minimax for --with-foundry implementer stacks — emits a model/minimax provider primitive instead of the default interface primitive
    • Global (global_en) and China (cn_zh) endpoint configuration in every generated stack
    • MiniMax-M3 (default) and MiniMax-M2.7 model options with context window, input modalities, thinking modes, and pricing
  • --region (global_en, cn_zh) and --model (MiniMax-M3, MiniMax-M2.7) flags to select the active MiniMax region and model
  • Help text and examples for the new flags
  • --with-memory memory-engineering bridge scaffold (tiers + budget signals)

[1.5.0] - 2026-07-20

Added

  • --with-foundry — one-command LE → harness-foundry funnel
    • Scaffolds .foundry/stack.yaml, hooks, sessions dir, and README
    • Pattern mapping: report-only → minimal, fix-capable → implementer
  • Post-scaffold Foundry CTA on every run (stronger when Loop Ready ≥ 80)
  • Help text for Foundry presets and examples

[1.4.0]

  • Prior release (starters, circuit breaker, intake, observability scaffolds)