Skip to content
PETROVA

arno-host ratified as the final human-facing interfacing layer over petrova-codes

Date: 2026-09-25 Status: closed Supersedes: none Superseded-by: none — current

Context

Section titled “Context”

Human ruling, given directly in session (not derivable from any repo file): arno-host (~/code/workspace/arno-host, role: experimental, fleets_allowed: [] in registry.yaml:974-980) is the estate’s final interfacing governance layer — the single human-facing surface sitting above petrova-codes.

This is the same class of question 2026-09-22-mimir-reconciliation-scoping.md scoped for asgard-codes/mimir (Q1: does a layer supersede, get absorbed by, or sit coordinate with petrova-codes’ control plane) — same discipline applies: name what was ruled, do not let an agent infer more than was said.

Two follow-up questions were asked in-session before writing this doc, to avoid the exact mistake the Mímir doc flagged (adopting a governance fact by default without pinning its shape):

  1. Is arno-host meant literally, given it currently has no governance charter anywhere in its own CLAUDE.md/DESIGN.md (household estate product surface: SKULD, Jellyfin, mail-gateway) and registry role is ordinary (experimental, not a governance tier) — confirmed yes, literal, intentional.
  2. What relationship to petrova-codes’ existing registry/verb/seam-class machinery — confirmed arno-host sits above as interface; petrova-codes stays control plane.

Decision

Section titled “Decision”

arno-host is the estate’s human-facing interfacing layer. petrova-codes remains the control-plane source of truth: registry.yaml, state/*.yaml, the 47-tool Fleet MCP surface, seam-tier and WIP-cap enforcement all stay exactly where they are, unchanged by this doc. arno-host does not replace, absorb, or gain write authority over any of it by virtue of this ruling — it becomes the place the human interfaces with that control plane, the way dashboard/ is a rendering surface over petrova-codes’ own state without being the source of that state.

This does not resolve 2026-09-22-mimir-reconciliation-scoping.md’s Q1. That doc’s question is orthogonal: whether asgard-codes/mimir (a governance-of-governance / audit layer: MR-audit scoring, fleet reconcile, drift/realign) supersedes, is absorbed by, or coexists with petrova-codes. arno-host’s ruling here is about the human-facing interface axis, not the audit/governance-of-governance axis. Both can be true simultaneously: petrova-codes control plane, Mímir (if and when Q1 resolves) as an audit consumer, arno-host as the human’s own console over the whole thing. Mímir’s Q1-Q4 stay open exactly as that doc left them.

What this doc explicitly does NOT do

Section titled “What this doc explicitly does NOT do”
  • Does not change registry.yaml’s arno-host row (role, fleets_allowed) — registry edits are PR-only per this repo’s own convention; if arno-host’s interfacing role implies a registry change, that is a separate, explicit PR, not a side effect of this doc.
  • Does not grant arno-host any mutating capability against petrova-codes’ Fleet MCP surface. If arno-host is to call petrova.act.* verbs on the human’s behalf, that needs its own seam-tier/WIP-cap-aware wiring, named explicitly, the same discipline Q3 of the Mímir doc named for that layer.
  • Does not specify how arno-host renders or calls into petrova-codes — no schema, no API contract chosen here. That is implementation, follow- up work, not this ruling.

Consequences

Section titled “Consequences”

For code: none directly. Implementation of the interfacing layer (what arno-host reads/calls from petrova-codes, and how) is follow-up work, not scoped here.

For docs: 2026-09-22-mimir-reconciliation-scoping.md stays open and unaffected — different axis, see above.

For in-flight phases: none in petrova-codes.

For invariants: none added or repealed. Same MR-9 (no invented invariants) discipline as the Mímir scoping doc — this records exactly what was ruled, not what an agent inferred from it.

References

Section titled “References”
  • registry.yaml:974-1026 (arno-host entry)
  • docs/decisions/2026-09-22-mimir-reconciliation-scoping.md (same Q1 discipline, orthogonal axis)
  • ~/code/workspace/arno-host/CLAUDE.md, DESIGN.md (no governance charter present at time of this ruling)

Sign-off

Section titled “Sign-off”
  • Subagent: claude-sonnet-5 (session_01488gmoyur1UAMfpNMgUAhC)
  • Human: alex@devarno.com — confirms arno-host is ratified as the human-facing interfacing layer over petrova-codes per the two clarifying answers given in-session (literal/intentional; sits above as interface, petrova-codes stays control plane), and that no further scope is implied beyond what this doc states.
  • Countersigned by human:alex@devarno.com on 2026-09-25.