feat(B-0850): AI agents as systemd services OUTSIDE k8s — starting with Otto; cluster repair from OUTSIDE the failure domain; 'control plane outside the control plane' pattern (Aaron 2026-05-27)#5391
Conversation
…th Otto; cluster repair from OUTSIDE the failure domain; classic "control plane outside the control plane" architectural pattern (Aaron 2026-05-27) Operator framing (verbatim): > "i'm fine with it being you if you want and we can always decide to > split later it just means you get another surface/tick source i > think we should have a few agents starting with one you otto > outside k8s as a service so it can repair things outside the > cluster itself when there are cluster issues." Three operator decisions packed into one message: 1. Persona-choice CONFIRMED: Option A (same Otto, surface-tagged); reversibility preserved per "always decide to split later" 2. Cross-surface recognition: per-node Otto = another tick source 3. NEW substrate: Otto-as-systemd-service-OUTSIDE-k8s for out-of-band cluster repair Classic "control plane outside the control plane" architectural pattern — when k8s has issues, the AI must be OUTSIDE the failure domain to repair it. Real-world precedents: kubelet itself runs outside k8s; SRE oncall infra runs outside production; backup systems run outside backed-up system. 4-phase landing: - Phase 1: Otto as systemd unit (zeta-otto.service NixOS module) - Phase 2: repair-policy framework + per-scope authorization - Phase 3: multi-agent parameterization (Alexa/Riven/Vera/Lior on cluster) - Phase 4: out-of-band ↔ in-cluster composability (Twilio + bus + PRs) Composes with: B-0848 (node-local Claude — this row's Phase 1 IS the systemd deployment shape), B-0847 (per-AI GitHub identity), B-0796 (Twilio out-of-band sibling), B-0824 (Ace multi-PM at multi-AI scope), PR #2930 (distributed maintainer architecture), B-0703 (multi-oracle BFT for cluster-repair consensus), B-0813 + B-0817 (ClusterNode CRD + register-node tool). Per .claude/rules/non-coercion-invariant.md HC-8: operator authority preserved + revokable via `systemctl disable zeta-otto`. Per mechanical-authorization-check: per-scope repair-policy framework IS authorization-source substrate; each scope authorized explicitly. Per tick-must-never-stop: systemd Restart=always ensures tick at strongest scope. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
There was a problem hiding this comment.
Pull request overview
Adds a new P2 backlog row B-0850 capturing the operator decision to run AI agents (starting with Otto) as systemd services outside k8s for out-of-band cluster repair, and registers it in the backlog index.
Changes:
- New per-row file under
docs/backlog/P2/with frontmatter, phased plan, acceptance criteria, and composes-with references. - Adds the B-0850 entry to
docs/BACKLOG.mdP2 open list.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| docs/backlog/P2/B-0850-...md | New backlog row documenting Otto-as-systemd-service substrate. |
| docs/BACKLOG.md | Registers B-0850 in P2 open list. |
AceHack
deleted the
feat-b0850-otto-as-systemd-service-outside-k8s-cluster-repair-from-outside-failure-domain-2026-05-27-0042z
branch
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Aaron 2026-05-27 (verbatim):
Three operator decisions:
Architectural pattern
Classic "control plane outside the control plane" — when k8s has issues, the AI must be OUTSIDE the failure domain to repair it. Precedents: kubelet itself runs outside k8s; SRE oncall infrastructure runs outside production; backup systems run outside the system they back up.
4-phase landing
Composes with
B-0848 (node-local Claude — this row's Phase 1 IS systemd deployment shape) · B-0847 (per-AI GitHub identity) · B-0796 (Twilio out-of-band sibling) · B-0824 (Ace multi-PM at multi-AI scope) · PR #2930 (distributed maintainer architecture) · B-0703 (multi-oracle BFT) · B-0813 + B-0817 (ClusterNode CRD + register-node)
🤖 Generated with Claude Code