npm update: loop-init --with-foundry + loop-audit 1.7 (LE → harness-foundry funnel)

[cobusgreyling]

One-command funnel: design the loop → version the harness

loop-engineering designs reliable inner loops. harness-foundry is the composable runtime in the middle. Today that path is one flag.

loop-engineering  →  harness-foundry  →  outerloop
   (patterns)         (runtime)          (governance)

Try it

npx @cobusgreyling/loop-init . --pattern daily-triage --tool grok --with-foundry

What you get:

1. Usual loop scaffold (skills, state, budget, constraints)
2. Loop Ready score from loop-audit
3. .foundry/stack.yaml mapped from your pattern
   • report-only (daily-triage, issue-triage, changelog-drafter) → minimal
   • fix-capable (ci-sweeper, pr-babysitter, …) → implementer
4. Next steps for foundry validate / foundry run

Without the flag, every loop-init run still prints a Foundry CTA — stronger when Loop Ready is ≥ 80.

Published packages

Package	Version	Highlights
@cobusgreyling/loop-init	1.5.0	--with-foundry
@cobusgreyling/loop-audit	1.7.0	Harness Runtime signals (.foundry/stack.yaml, sessions, outerloop emit, host)

Foundry side (aliases)

npx @cobusgreyling/harness-foundry init --from loop-engineering:daily-triage
# or: --from ci-sweeper  → implementer

Showcase: https://github.com/cobusgreyling/harness-foundry/blob/main/docs/showcase.md

PRs

• feat: LE → Foundry funnel (loop-audit 1.7 + loop-init --with-foundry) #324
• feat: LE pattern aliases for foundry init (funnel) harness-foundry#13

Ask

If you already have a Loop Ready badge, try --with-foundry (or re-run after a normal init) and share:

• Pattern you used
• Whether the stack preset felt right
• Anything confusing in the CTA / audit harness findings

Week-one culture unchanged: report-only before auto-fix. Foundry evolve proposals stay human-gated.

Thanks for building loops in public 🙏