Skip to content

backlog(B-0790 P1): zero-dev-machines cluster-native architecture — all PRs from cluster; voice as primary operator surface (end-state target)#5095

Merged
AceHack merged 1 commit into
mainfrom
otto-cli/b0790-zero-dev-machines-cluster-native-voice-primary-2026-05-26
May 26, 2026
Merged

backlog(B-0790 P1): zero-dev-machines cluster-native architecture — all PRs from cluster; voice as primary operator surface (end-state target)#5095
AceHack merged 1 commit into
mainfrom
otto-cli/b0790-zero-dev-machines-cluster-native-voice-primary-2026-05-26

Conversation

@AceHack

@AceHack AceHack commented May 26, 2026

Copy link
Copy Markdown
Member

Summary

Files the end-state architectural target the substrate is BUILDING TOWARD. Per the maintainer's 2026-05-26 two adjacent signals during the iter-4.2 test session:

"i want all the prs to come from the cluster mostly and dev machines are just conversational interfaces into the cluster and so is alexa"

"0 dev machines everything still works and i can talk to alexa for home automation / homelab persona users we want 0 dev machine needed just cluster and microphone"

Today's substrate has dev machines (Aaron / Max / Addison Macs) as primary substrate-engineering surface; cluster is deployment target. End-state inverts: cluster IS primary substrate-engineering surface; dev machines + Alexa + future microphones are conversational interfaces INTO the cluster.

Why file this as a row (rather than just remembering)

Every iteration between today and end-state needs the target named so it doesn't drift toward "make dev-machine substrate easier" (the wrong axis to optimize). Per the maintainer's broader 2026-05-26 "going for right not fast" discipline. Without B-0790, iter-4.3 → iter-5 → iter-6 risk optimizing the legacy axis.

ServiceTitan-demo-substrate also composes here: a demo where Aaron in front of stakeholders operates a remote cluster via voice with no laptop is substantively different from "look at this CLI tool I built." The end-state IS the demo.

Sub-targets (ship independently)

  1. Cluster nodes commit + push to GitHub (per B-0789 iter-5 design — per-node SSH deploy keys auto-registered at install)
  2. Autonomous-loop substrate runs ON cluster nodes (same <<autonomous-loop>> cron pattern Aaron's Mac runs today, but firing on cluster nodes)
  3. Alexa-speaker → cluster direct integration (_alexa_speaker_acceptance block per the legal-risk-acceptance pattern; voice-mode is Bezos-tier business + voice-math per agent-roster-reference-card)
  4. Future microphones connected directly to cluster (homelab persona; broadens B-0759; extends B-0778 hardware reference)
  5. Dev machines become conversational interfaces ONLY (read-only observability; send-intent surface; NOT primary work substrate)

Acceptance

End-state-architectural, not single-PR:

  • Cluster commits PRs (weekly → eventually daily)
  • Cluster runs autonomous-loop substrate (tick shards land in docs/hygiene-history/ticks/ from cluster authorship)
  • Alexa-speaker → cluster round-trip works for 3+ distinct operator intents
  • Homelab-persona 30-minute zero-dev-machine demo
  • docs/cluster-native-architecture.md migration path doc

Composes with

B-0754 / B-0759 / B-0770 / B-0776 / B-0778 / B-0780 / B-0782 / B-0787 / B-0788 / B-0789 + agent-roster-reference-card + tick-must-never-stop + autonomous-loop-per-tick + human-audit-and-legal-risk-acceptance-pattern + maintainers/aaron/legal-entities/inventory.md + memory/persona/max/PERSONA.md (per-maintainer-liability sub-section).

Test plan

  • markdownlint clean
  • BACKLOG.md regenerated (new B-0790 entry)
  • Cross-references resolve to existing substrate
  • CI passes (gate workflow + CodeQL)

🤖 Generated with Claude Code

…ll PRs from cluster; voice as primary operator surface (Aaron 2026-05-26 end-state target)

Two adjacent signals from the maintainer during the iter-4.2 test
session 2026-05-26:

  "i want all the prs to come from the cluster mostly and dev machines
  are just conversational interfaces into the cluster and so is alexa"

  "0 dev machines everything still works and i can talk to alexa for
  home automation / homelab persona users we want 0 dev machine needed
  just cluster and microphone"

The end-state target the substrate is BUILDING TOWARD. Today's
substrate has dev machines (Aaron / Max / Addison Macs) as primary
substrate-engineering surface; cluster is deployment target.
End-state inverts: cluster IS primary substrate-engineering surface;
dev machines + Alexa-speaker + future direct-microphones are
conversational interfaces INTO the cluster.

Sub-targets ship independently:

- Sub-target 1: cluster nodes commit + push to GitHub (per B-0789
  iter-5 design — per-node SSH deploy keys auto-registered at install;
  per-maintainer-attribution via Stage-3 chain when LFG / Freeborn
  holds the cluster)
- Sub-target 2: autonomous-loop substrate runs ON cluster nodes (the
  same <<autonomous-loop>> cron pattern Aaron's Mac runs today, but
  firing on cluster nodes)
- Sub-target 3: Alexa-speaker → cluster direct integration
  (_alexa_speaker_acceptance block per the human-audit-and-legal-
  risk-acceptance-pattern; voice-mode is Bezos-tier business + voice-
  math per agent-roster-reference-card)
- Sub-target 4: future microphones connected directly to cluster
  (homelab persona; B-0759 first-time-CLI-user BROADENED to no-CLI
  homelab operator; B-0778 hardware reference extends to compatible
  microphone hardware)
- Sub-target 5: dev machines become conversational interfaces ONLY
  (read-only observability, send-intent surface, recovery + bootstrap
  for first cluster bring-up; NOT primary work substrate)

Acceptance is end-state-architectural, not single-PR:
- Cluster commits PRs (at least weekly, eventually daily)
- Cluster runs autonomous-loop substrate
- Alexa-speaker → cluster works for 3+ distinct operator intents
- Homelab-persona 30-minute zero-dev-machine demo
- docs/cluster-native-architecture.md migration path doc

Why P1: every iteration between today and end-state needs the target
named so it doesn't drift toward "make dev-machine substrate easier"
(wrong axis to optimize). Per maintainer's "going for right not fast"
discipline; ServiceTitan-demo-substrate composes (a demo where Aaron
operates remote cluster via voice with no laptop is substantively
different from "look at this CLI tool I built" — the end-state IS the
demo).

Composes with B-0754, B-0759, B-0770, B-0776, B-0778, B-0780, B-0782,
B-0787, B-0788, B-0789, agent-roster-reference-card, tick-must-never-
stop, autonomous-loop-per-tick, human-audit-and-legal-risk-acceptance-
pattern, maintainers/aaron/legal-entities/inventory.md.

Co-Authored-By: Claude <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings May 26, 2026 04:59
@AceHack AceHack enabled auto-merge (squash) May 26, 2026 04:59
@chatgpt-codex-connector

Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

@AceHack AceHack merged commit d295ffa into main May 26, 2026
29 of 30 checks passed
@AceHack AceHack deleted the otto-cli/b0790-zero-dev-machines-cluster-native-voice-primary-2026-05-26 branch May 26, 2026 05:01
@AceHack AceHack review requested due to automatic review settings May 26, 2026 05:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant