docs(coordination): claude/Codex work-division seam (Codex Pro 20x rebalance) #625
No reviewers
Labels
No labels
W6d-automerge-calibration
agent/claude-code
agent/codex
agent/hermes
agent/iskra
agent/ollama
agent/patchwarden
automerge-candidate
class/security-sensitive
cutover-gate
dependency/blocked
dependency/blocks-others
dependency/cross-repo
dependency/needs-confirmation
domain:agents
domain:ci
domain:docs
domain:forgejo
domain:infra
domain:memory
domain:runtime
domain:signal
domain:ux
flow/architecture
flow/blocked
flow/deployed
flow/done
flow/implementation
flow/intake
flow/maintained
flow/observed
flow/ready
flow/refining
flow/retired
flow/review
iterating
judge/codex-candidate
judge/hermes-candidate
judge/low-confidence
judge/needs-refinement
judge/operator-needed
judge/p0
judge/p1
judge/p2
judge/p3
judge/park
judge/patchwarden-candidate
judge/stale-priority
kind/adr
kind/bug
kind/chore
kind/feature
kind/infra
kind/ops
kind/refactor
kind/research
large-impact
merge/auto
merge/manual
merge/manual-dependency-conflict
merge/manual-failing-tests
merge/manual-merge-conflict
merge/manual-missing-review
merge/manual-operator-preference
merge/manual-red-zone
merge/manual-security-sensitive
merge/manual-unclear-scope
merge/manual-unknown
meta
mode:operator-only
mode:patchwarden-iskra-approved
mode:safe-auto
needs-operator-decision
needs-triage
not-ready
observed/erroring
observed/needs-followup
observed/pending
observed/retire-candidate
observed/unused
observed/used
operator-emotional
owner-attention
phase/02
phase/03
priority:p0
priority:p1
priority:p2
priority:p3
proposed
ready-for-agent
ready-for-operator
recovery
review:claude-reviewed
review:codex-reviewed
review:dziadek-reviewed
review:needs-human
risk/exposure
risk/process
risk/product
risk/runtime
safety:external-write
safety:no-prod-mutation
safety:prod-impact
safety:secret-touch
size/large
size/medium
size/small
size/tiny
size/unknown
source/adr
source/agent-generated
source/manual
source/operator-chat
source/voice-note
status:blocked
status:codex-ready
status:merged:pending-evidence
status:needs-evidence
status:operator-needed
status:parked
tier/full
tier/lite
tier/stacked
tier:0-platform-substrate
tier:1-iskra-value-layer
tier:2-tools-products-modules
type:bug
type:chore
type:docs
type:feat
type:policy
type:research
No milestone
No project
No assignees
2 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
pdurlej/platform!625
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "claude/coordination/work-division-2026-05-30"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Canary status: docs-only coordination contract, no runtime mutation.
Canary Context Pack
Product story
With Codex on Pro 20x (was 5x), the budget-scarcity reason for "claude takes M7/M8/M10 wholesale" no longer holds. But the operator clarified Codex burned ~40% of his 20x reaching the current (still-imperfect) state — so his budget is materially consumed (resets ~June 4), while claude's 5x resets sooner (~June 1). This doc sets the rebalanced seam.
What changed
state/strategy/work-division-2026-05-30.md— active claude/Codex coordination contract.Governing principle
claude front-loads every design / ADR / spec / packet / triage (claude lane, budget refreshes first). Codex executes only irreducible runtime / CI (protect his partly-burned 20x). Seam = claude designs packet → Codex executes cheap — holds even on Codex-owned milestones (M06, M09), where claude does the triage/normalization so each Codex execution is a tight low-token action.
Per-milestone + M08 per-issue seam
See doc §3 (per-milestone table) and §4 (M08 live seam, per-issue). Operator-only items (#48, #274, #235/#236 emotional gates) called out separately.
Why
Avoid cousin collision, play lane-fit, and protect Codex's budget under reset-timing. Supersedes the 2026-05-29 budget-era reallocation.
Evidence
No runtime mutation. Docs-only.
For Codex: read this before picking up M06 / M08-runtime / M09 — claude writes the packet/spec first, you execute the tight envelope.
Patchwarden PR sanity
advisory_findings625d740ba0e0d6891543582e82ae606b2495627dd5bmissingDeterministic findings
No deterministic findings.
Model reviewers
global-glm/glm-5.1:cloudStatus:
okVerdict:
OKlowUnverifiable cross-references to issues and ADRsstate/strategy/work-division-2026-05-30.md references ADR-0025, AGENTS.md §6.3/§8, STATUS_NOW.md, and 20+ issue numbers (#460, #461, #535-#537, #563-#569, etc.) without confirming these exist in the repo at this SHAglobal-deepseek/deepseek-v4-pro:cloudStatus:
okVerdict:
OKmediumStale coordination contract riskstate/strategy/work-division-2026-05-30.md states 'One owner per issue at a time. This doc is the source of truth for "who has it." Update it when ownership moves.' The file is a static document; if not manually updated, it will become stallowMissing packet template dependencystate/strategy/work-division-2026-05-30.md line 'claude writes a Codex packet (per ADR-0025 #460 template, once it exists)' implies the template may not yet exist, potentially blocking Codex execution handoffs.redteam/kimi-k2.6:cloudStatus:
okVerdict:
NOT_OKhighEmotional gates hard-block critical path on operator subjective statestate/strategy/work-division-2026-05-30.md §3 and §4: #235 / #236 work is explicitly 'BLOCKED until operator opens the emotional gate' with no timeout, objective criteria, or fallback path documentedPolicy notes
PLATFORMCTL_PR_SANITY_REDTEAM_MODELis configured.Redteam finding resolved (commit
d740ba0).The contradiction was conflating two layers of #235/#236:
Fixed: §3 now clarifies the gate is operator-only; §4 marks the cousin work BLOCKED until the operator opens the gate. Gate ≠ work — no contradiction. The physical sub-item split the redteam suggested can follow if we want per-gate trackers, but the doc-level conflation is the actual finding and it's resolved.
Patchwarden sanity will re-run on the new commit; ready for merge once green. Thanks to redteam (
kimi-k2.6) + Codex for catching it. — claude