prompts(codex): dispatch for F2 ramp continuation (5 stateless smokes) #280

Merged

pdurlej merged 1 commit from claude/f2-ramp-continuation-dispatch into main

2026-05-14 10:58:24 +02:00

claude commented

2026-05-14 10:57:13 +02:00

Collaborator

Summary

F2 ramp continuation after re-audit. Pan Herbatka's original 7-service F2 list shrunk to 5 truly stateless + strict-v2-ready candidates after checking spec.runtime.statefulness:

mirotalk-sfu, mirotalk-admin (F1.5 leftovers; PRs #258/#259 merged)
umami (analytics)
n8n-worker (worker is stateless even though n8n-main isn't)
playwright-mcp (peripheral MCP)

Deferred (intentional)

honcho-api, honcho-deriver, coredns — defer to joint session (Iskra/DNS blast radius)
karakeep, matrix-hub-private, sidecar-bound services — special procedure
agaria-* — separate project, outside cutover
np-meerkat-frontend — RED yesterday → #269 Medium PR
13 stateless-but-not-strict-v2 services — strict-v2 fill PR first, lower priority

Natural co-benefit

F2 smokes trigger workflow_dispatch, which gives Codex's #278 watchdog instrumentation (live on RS2000 release-root f592fc5...) real observation data for #260. Codex's own recommendation post-RCA: "czekamy na następny realny stuck event z aktywną diagnostyką #278". F2 ramp IS that next observation window.

Tier

Trivial per ADR-0007 (prompt file, no runtime/schema/sacred-path mutation).

Operator action

Merge → paste contents of prompts/codex-f2-ramp-continuation-2026-05-14.md to fresh Codex session. Or fire after current Codex session finishes.

Refs #142 #260 #269 #279

## Summary F2 ramp continuation after re-audit. Pan Herbatka's original 7-service F2 list shrunk to **5 truly stateless + strict-v2-ready** candidates after checking `spec.runtime.statefulness`: - `mirotalk-sfu`, `mirotalk-admin` (F1.5 leftovers; PRs #258/#259 merged) - `umami` (analytics) - `n8n-worker` (worker is stateless even though `n8n-main` isn't) - `playwright-mcp` (peripheral MCP) ## Deferred (intentional) - `honcho-api`, `honcho-deriver`, `coredns` — defer to joint session (Iskra/DNS blast radius) - `karakeep`, `matrix-hub-private`, sidecar-bound services — special procedure - `agaria-*` — separate project, outside cutover - `np-meerkat-frontend` — RED yesterday → #269 Medium PR - 13 stateless-but-not-strict-v2 services — strict-v2 fill PR first, lower priority ## Natural co-benefit F2 smokes trigger `workflow_dispatch`, which gives Codex's #278 watchdog instrumentation (live on RS2000 release-root `f592fc5...`) **real observation data** for #260. Codex's own recommendation post-RCA: *"czekamy na następny realny stuck event z aktywną diagnostyką #278"*. F2 ramp IS that next observation window. ## Tier Trivial per ADR-0007 (prompt file, no runtime/schema/sacred-path mutation). ## Operator action Merge → paste contents of `prompts/codex-f2-ramp-continuation-2026-05-14.md` to fresh Codex session. Or fire after current Codex session finishes. Refs #142 #260 #269 #279

claude added 1 commit

2026-05-14 10:57:13 +02:00

prompts(codex): dispatch for F2 ramp continuation (5 stateless smokes)

base-is-main / guard (pull_request) Successful in 1s

Details

canary-required / collect-diff (pull_request) Successful in 4s

Details

canary-required / canary (pull_request) Successful in 13s

Details

81681c932d

Re-audited F2 candidates against spec.runtime.statefulness — only 5
stateless + strict-v2-ready remain: mirotalk-sfu/admin, umami,
n8n-worker, playwright-mcp. Honcho-api/deriver/coredns deferred to
joint session (Iskra/DNS blast radius). Karakeep + sidecar-bound
deferred (special procedure). agaria-* outside cutover scope.

F2 smokes will naturally exercise the #278 watchdog instrumentation
landed on RS2000 release-root f592fc5..., providing real-evidence
window for #260 stuck-pickup hypothesis without manufacturing tests.

Hard gates G4-G7 preserved. Lite-tier fix PRs allowed only if smoke
uncovers fixable issue; cross-references #279 release-root gap.

Refs #142 #260 #269 #279