oduffy-delphi/workstream-complete

Workstream Complete — Wrap Up Completed Work

Close out a finished vein of work: capture lessons and update documentation to reflect completion. No handoff — this is for work that's done, not being passed forward.

/workstream-complete and /handoff are mutually exclusive. This caps a workstream; /handoff passes one on. In-flight work → STOP and invoke /handoff instead. Two workstreams (one done, one in-flight) → end each separately, naming which is which. → coordinator CLAUDE.md § Handoff Lineage.

Instructions

Capture lessons and update plan/project documentation to reflect completion. If work is incomplete, use /handoff instead. Multiple agents may be running concurrently — this skill closes out ONE session without heavy repo-wide operations.

Execution Shape — gates vs. todo-list

This skill is mirror-shaped to /handoff: a small set of sequential gates plus a TODO-LIST cluster of independent post-work cleanup steps. Treat them as such — do not ladder-walk the todo-list. Convention: docs/wiki/skill-step-parallelization.md.

Sequential gates (real data-dependency edges — must be in this order):

1. Step 1 → Step 1.2 micro-chain — classification reads the lesson Step 1 just wrote. Skip both together if no new lesson.
2. Step 2.6 internal chain (Steps 2.6.1 → 2.6.2 → 2.6.3 → 2.6.4 → 2.6.5 → 2.6.5a → 2.6.6 → 2.6.7) — the per-entry archive write is a real chain: AUTO-MIGRATE → chain-slug resolve → Sonnet nature-infer → session-id resolve → LoE block → write entry. Internal to Step 2.6 only.
3. Step 2.9 (code review) — integrator-edited files must be staged in Step 3.
4. Step 2.4 → Step 3 staging edge — when a governing plan exists, Step 2.4's reconciled plan doc (the corrected docs/plans/<feature>.md) must be staged in Step 3 before commit. Step 2.4 is a micro-chain off Step 2 in the todo-list cluster (see below); its output is part of Step 3's fan-in union, identical in shape to the existing Step 2.9-integrator-edits → Step 3 staging edge.
5. Step 3 (commit + verify remote) — fan-in of ALL preceding file edits (lessons, plan docs, archive entries, orientation cache, action-items, review-integrator outputs, reconciled plan doc); commit consumes the union via explicit-path staging.
6. Step 3.5 (archive session claim) — consumes Step 3's pushed commit.
7. Step 4 (final summary) — informational.

Todo-list (execute in any order, batch parallel where two independently read/write different files — with the Step 2→2.4 micro-chain exception noted below):

• Step 1 (then 1.2) — run as an inseparable pair, one todo-list slot — lessons capture + classification (state/lessons.md). The 1→1.2 edge is real; run them sequentially as a unit; the pair parallelizes with the other slots.
• Step 2 (then 2.4) — run as a micro-chain, one todo-list slot — plan documentation (docs/plans/, tasks/<feature>/todo.md, etc.), then plan-doc reconciliation. The 2→2.4 edge is real: Step 2.4 reconciles the plan doc Step 2 just updated. Run them sequentially as a unit; the pair parallelizes with the other slots. Skip Step 2.4 if no governing plan exists.
• Step 2.5 — doc-alignment insurance (chunk/stub **Status:** fields)
• Step 2.6 — archive uncaptured work (archive/completed/YYYY-MM/; internal chain 2.6.1→2.6.7 is real but isolated to this slot)
• Step 2.7 — archive predecessor handoff (file move only — independent of all other slots)
• Step 2.8 — refresh orientation documents (pinboard + tracker + action-items + docs README)
• Step 2.95 — aftermath lens-checkers (big-workstream only; reads session context, writes only the Step 4 summary checklist) — parallel-safe with the 2.x cluster

These six slots touch disjoint surfaces. Among peer slots, none consumes another's output — where two slots operate on different paths, run them in the same response via parallel tool calls. Step 3 is a fan-in: it stages the union of all files touched by the cluster; peer ordering is irrelevant, only their position before Step 3 matters.

Step 1: Capture Lessons

Read state/lessons.md (if it exists). If anything was learned this session that isn't already captured, add it — but apply the intake filter first.

Create on first use: If lessons exist to capture AND the file does not exist yet, create it with a # Lessons — [Project Name] header, a one-line purpose note, and the <!-- EM-maintained; see CLAUDE.md § Self-Improvement Loop --> comment, then append the entry. If no lessons to capture and the file doesn't exist, do not create it.

Feature scope: <feature> is derived from the current work context:

• If a feature-scoped plan exists at tasks/<feature>/todo.md, use that feature name
• If on a feature/<name> branch, use <name>
• Otherwise, use state/lessons.md (global)

Qualifies: user corrections, surprising API/tooling behavior, patterns that worked or failed, debugging insights. Doesn't qualify: one-off fixes, pipeline-run details, anything already in code/CLAUDE.md/MEMORY.md. Test: "Will this save time in the next 4 weeks?"

Format: bold title + 1-2 sentence rule, max 3 lines. Prefer merging with an existing entry. Skip if nothing new.

Step 1.2: Lesson Classification

For each new lesson, ask: "Would this apply to any project type using the coordinator pipeline?" Autonomous self-classification; no review step.

• Yes (universal): (a) tag with [universal] on the bold title line; (b) append one-liner to ~/.claude/state/coordinator-improvement-queue.md: - YYYY-MM-DD | <source-repo> | state/lessons.md:<line> | <summary> | proposed target: <coordinator file>. Skip if that <source-file>:<line> already exists.
• No (project-specific): no further action.
• Nothing new in Step 1: skip entirely.

Step 2: Update Plan Documentation

Find and update relevant plan/task documentation to reflect what was completed:

1. Find the plan docs — actively search, don't wait to recall. Check in order: session context (opened docs), tasks/<feature>/todo.md, tasks/plans/, docs/plans/, ~/.claude/plans/, tasks/todo.md/tasks/plan.md. If a plan exists for work this session touched, read and update it — sessions diving from handoffs often never explicitly opened the plan.
2. Mark completed items: Check off finished tasks, update status fields, add completion notes where appropriate.
3. Add a review section (if not already present) summarizing outcomes — what was built, key decisions, anything notable about the result.
4. Update other pertinent docs: If the work affected README files, architecture docs, or other project documentation that should reflect the new state, update those too. Use judgment — only touch docs that are clearly stale as a result of this session's work.

Step 2.4: Reconcile Plan Doc Against Shipped Reality

Spec backlink: archive/specs/2026-05-26-session-end-deviation-reconciliation-gate.md § Goal, D1–D5.

Governing-plan predicate — fires only when a governing plan/spec exists (docs/plans/YYYY-MM-DD-<feature>.md, RFC, enriched stub, or handoff-body-as-live-spec). Negative-spec: if no governing plan exists, skip entirely. Do NOT invent a plan to reconcile against. No ceremony tax on plan-less sessions. → docs/wiki/ceremony-calibration.md.

What to correct in place — ALLOWLIST sections

Plan documents contain sections that /distill crystallizes into wiki entries (the ALLOWLIST). When what shipped diverged from the plan's forecast, correct these sections in the plan doc before the Step 3 commit so distill synthesizes shipped reality, not the stale forecast:

• Decisions Made — if a decision record describes an approach that was modified or superseded during implementation, annotate the changed item: SHIPPED: <what-shipped> (was: <plan-forecast>). For decisions that shipped unchanged, no annotation is needed.
• API Contracts / Function Signatures — if a function signature, interface shape, or protocol contract landed differently from the plan's specification, correct the relevant entry with the same annotation: SHIPPED: <actual-signature> (was: <planned-signature>).
• Acceptance Criteria oracle tables — AC tables (columns: ID | Criterion | Test | Binding-Class | Status) are consumed by check-acceptance-oracle.sh. In-place correction of an AC table is scoped to the Status/note columns only (e.g. Status → shipped-differently with a note cell). Free-text mutation of the Criterion or Test cells would corrupt the structured oracle the parser expects. Substantive "what shipped vs forecast" delta routes to the ## Deviations table (below) and to the Decisions Made section — NOT a free-text edit of Criterion/Test cells.

The (was: <plan-forecast>) annotation maps to distill's [SUPERSEDED] nugget class — superseded provenance only; what crystallizes is the shipped reality.

Append the ## Deviations audit section

After correcting ALLOWLIST sections, append the following section to the plan doc (or extend it if already present from an earlier partial session):

## Deviations

| deviation | reason | commit |
|-----------|--------|--------|
| <brief description of what diverged from the plan> | <why it diverged> | <commit sha or "pending"> |

One row per meaningful deviation. This section is provenance-only — it is not crystallized by /distill (it is dropped as [EPHEMERAL]). The essential deviation fact survives in the corrected ALLOWLIST sections' (was: …) annotations; the verbose reasons live here and in git history.

Soft-ordering note re Step 2.9

Step 2.9's spec-completion lens is a soft input to this step, not a hard predecessor — Step 2.4 reconciles what the EM knows from session context. If Step 2.9 surfaces additional drift, fold those findings before Step 3. Two write-back types: Step 2.4 performs forecast→reality corrections; Step 2.9's integrator path performs defect-fix write-backs. Both fan into Step 3's staging union.

Step 2.5: Doc-Alignment Insurance

End of session is the last chance to ensure status fields match reality. This catches work that completed but whose status wasn't updated — common after compaction or rapid context shifts.

1. Check active chunk/stub docs: If this session worked on chunk stubs (files with **Status:** fields in docs/active/, docs/plans/, or similar), verify their status reflects what actually happened:
   • If the work is complete but status says "in progress" → update to complete
   • If the work is blocked but status says "in progress" → update to blocked with reason
   • If the status is already correct, skip
2. Check execution tracker: If a tactical execution tracker exists (e.g., docs/plans/consolidated-execution-tracker.md), verify that chunks worked on this session have accurate status entries
3. Lightweight pass only. Read what's in your conversation context — don't re-read every file in the project. If you have no memory of working on tracked chunks, skip this step entirely.

Step 2.6: Archive Uncaptured Work

Sweep the session's commits for completed work that isn't already in the project tracker (docs/project-tracker.md) or the per-entry completion archive under archive/completed/. This catches bug fixes, ad-hoc requests, and quick tasks that bypassed the spec pipeline.

Skip if no archive/ directory exists and no docs/project-tracker.md exists — the project hasn't adopted unified tracking yet.

Step 2.6.1 — Scan session commits

git log --oneline for commits since session start. For each substantive commit (skip merges, doc-only, quick-saves), check if already represented in the tracker or archive/completed/YYYY-MM/ (by SHA). Group related commits into one entry.

Step 2.6.2 — AUTO-MIGRATE legacy monolith (idempotent)

Before writing any per-entry file, check whether a legacy monolith file exists at archive/completed/YYYY-MM.md (i.e., directly at the root of archive/completed/, NOT under a YYYY-MM/ subdirectory). If found AND COORDINATOR_OVERRIDE_LEGACY_MONOLITH is not set to 1:

git mv archive/completed/YYYY-MM.md archive/completed/legacy/YYYY-MM.md

Create archive/completed/legacy/ if it does not exist. The git mv is idempotent — subsequent runs find no monolith-at-root and skip silently. If COORDINATOR_OVERRIDE_LEGACY_MONOLITH=1, skip the git mv (the EM has already handled migration manually).

<!-- TRIPWIRE: NO monolithic append — archive/completed/YYYY-MM.md writes outside legacy/ are forbidden. Static-grep check: check-no-monolith-completion-append.sh (created in Chunk 10). Registered in docs/wiki/coordinator-tripwires.md. Override: COORDINATOR_OVERRIDE_LEGACY_MONOLITH=1 skips git mv (manual migration path). -->

Step 2.6.3 — Determine chain slug

(a) If a plan was touched this session (any file under docs/plans/ or tasks/*/todo.md), chain = that plan's filename stem (e.g., 2026-05-19-completion-log-phase1). (b) Else if a handoff was picked up this session, chain = the handoff's filename stem. (c) Else if a workstream slug appears in any handoff frontmatter consumed this session, chain = that slug. (d) Else chain = null (omit from filename; write as archive/completed/YYYY-MM/YYYY-MM-DD-adhoc-<sid6>.md).

Step 2.6.4 — AUTO-INFER nature via Sonnet dispatch

Nature is classified automatically — no interactive prompt. Dispatch a small Sonnet sub-call (~1 KB output) with:

• Touched paths (from git diff --name-only for this session's commits)
• Commit messages (from git log --oneline for this session)
• Workstream kind (plan-driven | handoff-pickup | spinoff | ad-hoc)
• Chain slug (resolved in Step 2.6.3)

Sonnet classifies to one of [roadmap | bugfix | tech-debt | infra] and returns a nature: value + one-sentence rationale. Tag the entry nature_inferred: true.

Interactive override: If COMPLETION_NATURE is set in the environment before invoking /workstream-complete, use that value as nature: directly and write nature_inferred: false. The env var bypasses the Sonnet dispatch entirely.

Why AUTO-INFER not interactive-prompt: workstream-complete fires in autonomous chains where no human is present; a skip-default would bias --where nature=<x> queries. See plan § Chunk 3 for full rationale.

Step 2.6.5 — Resolve $em_sid and derive <sid6>

$em_sid sourcing (env-var-primary):

1. If $em_sid is set in the environment, use it directly.
2. Else use $CLAUDE_CODE_SESSION_ID — the platform-injected session id (Claude Code ≥ ~2.1.150). Per-session and unclobberable by a sibling session, so it is authoritative.
3. Else read from .git/coordinator-sessions/.current-session-id (last-writer-wins sentinel — session-init.sh writes it on every SessionStart; only a fallback for old Claude Code, per docs/wiki/claude-code-platform-gotchas.md). If the sentinel read is ambiguous (flips between ids across reads), two sessions are live — do not trust it; the env var in step 2 is the answer.
4. Do NOT use meta.json-based lookup — it is circular (you need $em_sid to find the directory containing meta.json).

<sid6> = last 6 characters of the resolved $em_sid. If $em_sid cannot be resolved, generate a 6-char hex fallback from the current timestamp (date +%s | tail -c 7 | head -c 6).

The <sid6> suffix ensures uniqueness per session with no race condition.

Step 2.6.5a — Compute LoE block

Behavioral rule (tripwire): Workstream-complete MUST invoke coordinator-session-loe.sh (or aggregate-chain-loe.sh for chain-terminal) to write per-session LoE into the completion entry. Skipping this step produces an incomplete entry that Phase 3 consumers and workweek-complete cannot query. No override mechanism; the loe: block is always written.

Determine whether this is a single-session or chain-terminal session using the same detection logic as Step 2.9 chain-end detection:

• Single-session: this session was NOT opened via /pickup (no predecessor handoff consumed).
• Chain-terminal: session was opened via /pickup AND is ending via /workstream-complete (not /handoff or /spinoff).

Single-session path:

loe_block=$(~/.claude/plugins/coordinator/bin/coordinator-session-loe.sh \
  --format yaml-frontmatter 2>/dev/null)

If the script is absent or returns non-zero, degrade gracefully: set loe_block to:

loe:
  agent_dispatches: null
  opus_dispatches: null
  em_tokens: null
  tshirt: null

Chain-terminal path:

Resolve the consumed predecessor handoff path (the handoff archived by Step 2.7 this session). Resolution order:

1. Check session state for the handoff path that was consumed at /pickup time.
2. Walk state/handoffs/archive/<YYYY-MM>/ for entries with consumed_by: <this session_id>.

Then invoke the chain aggregator:

loe_block=$(~/.claude/plugins/coordinator/bin/aggregate-chain-loe.sh \
  --terminal-handoff "<resolved-predecessor-path>" \
  --format yaml-frontmatter 2>/dev/null)

If the script is absent or returns non-zero, degrade the same way as the single-session path. The resolved $loe_block is embedded into the completion entry frontmatter in Step 2.6.6.

Step 2.6.6 — Write per-entry file

For each untracked completed work item (or one entry if work is cohesive), write:

archive/completed/YYYY-MM/YYYY-MM-DD-<chain-slug>-<sid6>.md

(chain null → YYYY-MM-DD-adhoc-<sid6>.md). Create the YYYY-MM/ subdirectory if absent.

File shape:

---
title: "<Concise past-tense one-line description>"
created: YYYY-MM-DD
nature: <roadmap|bugfix|tech-debt|infra>
nature_inferred: <true|false>
chain: <chain-slug or null>
commits:
  - <sha1>
  - <sha2>
status: pending-release
chain_terminal: <true|false>
authored_by: <em_sid or null>
loe:
  agent_dispatches: <N or null>
  opus_dispatches: <N or null>
  em_tokens: <N or null>
  tshirt: <XS|S|M|L|XL|null>
# chain-terminal only — omit for single-session entries:
# chain_sessions: <N>
# chain_span_days: <N>
# chain_starting_handoff: <path>
---

<One paragraph prose summary: what shipped, key decisions, anything notable.>

Frontmatter semantics: nature_inferred true on AUTO-INFER, false on COMPLETION_NATURE env override; chain_terminal true on /pickup→/workstream-complete; authored_by = $em_sid or null; status: pending-release always; loe from Step 2.6.5a (chain-terminal adds aggregate + chain-summary fields; loe.tshirt: null = script unavailable). <sid6> ensures uniqueness — no collision handling needed.

Step 2.6.7 — Judgment filter

Not every commit is a work item. Group related commits into a single archive entry. Skip trivial commits (typo fixes, formatting). If a session produced no substantive commits beyond doc/lesson housekeeping, no archive entry is needed — skip silently.

Step 2.65: Cross-repo memo lifecycle sweep — flip resolved memos to actioned

When the session's work resolves a cross-repo memo in this repo's cross-repo/inbox/, flip status: open → actioned (with optional decision: line) so the inbox accurately reflects channel state.

Detection — non-automatable; prompt the EM (no reliable programmatic signal connects commits to memo resolution):

1. Glob cross-repo/inbox/*.md in the current repo. Parse YAML frontmatter; filter to status: open (or absent → treat as open).
2. If zero matches: skip this step silently.
3. If ≥1 matches: list each as a numbered line — N. <basename> — <title or first heading> (from: <from>, topic: <topic>). Then ask once, plain prose: "Any of these resolved this session? If yes, give me the numbers; I'll flip status: actioned and add a one-line decision: you dictate. (Type none if none.)"
4. For each named memo: Edit the file in place — set status: actioned and append decision: <PM-supplied line> to the frontmatter. Then sweep it out of the inbox: git mv cross-repo/inbox/<file> cross-repo/archive/<file>. Create cross-repo/archive/ if absent. Both the edit and the move fold into Step 3's commit.

Out of scope here: sender side; memos the session created or moved; memos in cross-repo/archive/. Do NOT touch any other repo's cross-repo/.

Why here: batching at session close is cheaper than inline at resolution time.

Step 2.66: Sender-side — do NOT re-surface already-sent memos

If this session sent a cross-repo-memo or doctrine-seeded a sibling repo, do NOT list it as "pending PM-relay" or "pending your action" in the Final Summary, Flag to PM:, or any follow-on /handoff. The receiver's inbox is the canonical channel; sender-side status knowledge decays. Banned phrasings: "PM-relays pending your action", "Cross-repo memo X awaiting relay", "doctrine-seed Y pending sibling-EM action." → docs/wiki/cross-repo-communication.md § Don't re-nag the PM about already-sent memos.

Step 2.7: Archive Predecessor Handoff (if applicable)

When this session was opened with /pickup, the consumed handoff still lives in state/handoffs/ (mutation-only at pickup time). If this session is ending via /workstream-complete rather than /handoff, archive the predecessor now.

Detection: scan state/handoffs/*.md, read frontmatter consumed_by:. Resolve session id: $CLAUDE_CODE_SESSION_ID first; .git/coordinator-sessions/.current-session-id fallback. Zero matches → skip. One match → archive. Multiple matches → log to stderr and archive all.

Action: git mv state/handoffs/<file> archive/handoffs/<file>. Create archive/handoffs/ if absent. On git mv failure (already moved by a concurrent /handoff), log to stderr and continue. The move folds into the Step 3 commit.

No claim release call needed — cs_archive at Step 3.5 carries the entire session directory (including handoff-claims/) into .archive/. Skip entirely if exiting via /handoff — the two paths are mutually exclusive.

Step 2.75: Refresh Handoff Tracker

After Step 2.7, regenerate state/handoff-tracker.md:

node ~/.claude/plugins/coordinator/bin/render-handoff-tracker.js

Skip silently if the script is absent or fails. Stage state/handoff-tracker.md in Step 3's scoped commit.

Step 2.8: Refresh Orientation Documents

Update the documents that future sessions read for orientation — closing the read-write loop with /workstream-start and /workday-start.

1. Orientation cache (state/orientation_cache.md): Do not author the cache body. Do not patch sections. Do not re-derive content section-by-section. The cache schema (pipelines/workday-start-internals.md § 5.5) is owned by ceremony writers (/workday-start, /update-docs). /workstream-complete is a mid-session writer with a single, narrowly-scoped capability: pinboard append.

   Pinboard rule (the only cache mutation permitted here): if this session surfaced something the next session start MUST see, and it would otherwise be lost (a transient surface gotcha; a critical blocker context; an environment-specific caveat that fooled this session and will fool the next), write exactly one line to ## Pinboard via the routine:

   bash ~/.claude/plugins/coordinator/bin/regenerate-orientation-cache.sh \
       --invoker workstream-complete \
       --pinboard "YYYY-MM-DD <writer-slug>: <one-line note>"
   

   One-slot escape valve — a second write overwrites, not appends. Cleared at every ceremony regen. If you want to write more than one line, that's a wiki edit, handoff body, or lessons.md entry. If nothing pinboard-worthy, do nothing. If the cache file doesn't exist (ls state/orientation_cache.md before asserting), skip.

2. Project tracker (docs/project-tracker.md): If it exists and this session completed or progressed tracked items, update their status rows. Only touch rows this session affected — don't re-derive the whole tracker.

3. Action items (first match: ACTION-ITEMS.md, docs/active/ACTION-ITEMS.md, docs/ACTION-ITEMS.md): If one exists and this session resolved any listed items, check them off or remove them per the file's existing conventions.

4. Documentation index (docs/README.md): If it exists and this session created new guides, added research files, or completed plan documents, patch the relevant table. Only touch rows this session affected.

Concurrency note: Targeted patches only — safe with concurrent agents working on different items.

Step 2.9: Code Review Consideration

Assess whether this session's diff warrants a code review pass before committing. EM makes the call using the table below — this step is judgment, not ceremony.

Diff-shape table:

| Session shape | Default scale | |---|---| | Doc-only edits, lesson capture, no executor dispatched, no code touched | None | | Single-file fix <50 LOC, no shared schema touched, no executor | None (but commit message names the change) | | Any executor dispatched, OR >50 LOC code change, OR shared schema/seam touched | code-reviewer (Sonnet, locked — see agents/code-reviewer.md) | | Big-diff brightline (any one of: ≥500 gross LOC (insertions+deletions), OR ≥5 commits, OR ≥4 distinct surfaces — e.g. bash + JSON + tests + doctrine. File count is reported for context but is NOT a gate; mass-renames touch many files at zero review-cost.) | Partitioned code-reviewer dispatches — mandatory, not chain-end-gated. See § Partitioning large surfaces | | Chain-end (started with /pickup, ending without /handoff//spinoff) AND chain diff is non-trivial | code-reviewer on chain diff | | Chain-end AND chain diff exceeds the big-diff brightline above | Partitioned code-reviewer dispatches. Named reviewers (the Staff Engineer, personas) are for plans and architecture, not code output. Sonnet code-reviewer is the ceiling at workstream-complete |

Precedence rule: the big-diff brightline (row 4) and chain-end rows (5, 6) override workstream-complete rows (1, 2, 3) when they apply — partitioning is the integration-risk control, not a chain-end privilege.

Brightline gate — mechanical, before picking a row. Run bin/review-brightline-gate.sh [<range>]; verdict PARTITION-MANDATORY overrides row choice. Single-reviewer above the brightline is a doctrine violation — wiki § Worked counterexample.

Anchored-ranges note: the small-side anchor (50 LOC) is calibration — shape can adjust. Big-side brightlines (≥500 gross LOC / ≥5 commits / ≥4 surfaces) are hard floors. (Recalibrated 2026-06-09 — files count dropped as a gate, commits added, surfaces bumped 3→4. Rationale: file count is blunt — mass-renames touch many files at zero review-cost — while commit count tracks independent logical slices, which is what slicing operates on; 3 surfaces fired on routine hook-fixes (shell+test+wiki), 4 demands genuine breadth.)

Partitioning large surfaces across multiple code-reviewer dispatches (rows 4 and 6 — the two partition-mandatory rows): Fan out into parallel code-reviewer dispatches over coherent slices (by package boundary, concern, or directory cluster) — no lens-orthogonality manifest or synthesizer required. Mechanics:

1. Each code-reviewer prompt names its slice explicitly (paths or commit subset) and an "out of scope: the rest of the chain diff" line.
2. Dispatch in parallel; integrators are 1:1 with reviewer slices — one coordinator:review-integrator per code-reviewer slice, dispatched in parallel, each scoped to the same slice paths as its source reviewer. No collation into a single union-integrator. Mechanism: bin/fan-out-integrator.sh (input: TSV of <slice-id>TAB<reviewer-sidecar-path>TAB<scope-files>; output: N parallel coordinator:review-integrator dispatch blocks). Manual N-prompt construction is permitted only when the script is unavailable — collation is never permitted. The reasoning is structural: reviewers were partitioned because one Sonnet couldn't fit the whole surface; the same context-fit constraint binds the integrator. A union-integrator inherits N reviewers' findings against N disjoint file sets and the merged scope is exactly what the slicing avoided — see docs/wiki/review-integration-doctrine.md § Integrator dispatches are 1:1 with reviewer slices.
3. Trail write uses --reviewer code-reviewer; record partition shape in the wrap-up sentence.
4. Post-review the Staff Engineer-escalation criteria apply to the combined finding set. No upper bound on partition count — the constraint is per-reviewer context fit, which (per item 2) is also the constraint on per-integrator scope.

No named-reviewer escalation from code review. Code output review is Sonnet code-reviewer only — partition across slices as needed. Architectural findings from code-reviewer → state/lessons.md + surface to PM; do not escalate to a named reviewer within the code-review path.

The weekly /workweek-complete Step 7 merge-gate is a separate, independent ceremony — do not skip workstream-complete review and "surface to PM for workweek."

Anti-ceremony-bias tripwire: code-reviewer is the floor on row-3+ sessions, not a negotiable add-on. Drafting a "waive with rationale" sentence on a row-3+ session is the tell — run the review. EM keeps waive authority on genuinely shallow row-3 diffs; the test is diff shape, not row number. → docs/wiki/workstream-complete-review.md § Why post-implementation review is not redundant with plan-time review.

Chain-end detection:

• Resolve session-id: $CLAUDE_CODE_SESSION_ID env var first (platform-injected, unclobberable); .git/coordinator-sessions/.current-session-id sentinel fallback only when the env var is empty.
• Chain-end signal: session opened via /pickup AND ending without /handoff or /spinoff invocation this session.
• Trail is the only valid code-output coverage signal. Read records via list-review-trail-records.sh (live AND archived — live-only glob misses prior-week reviews). Plan-level the Staff Engineer reviews are NOT trail records; a "the Staff Engineer reviewed the plan" handoff note does NOT satisfy the chain-end code-reviewer floor. No trail record covering the sha-range = unreviewed.

Diff scope:

• Chain-end → git log $(git merge-base origin/main HEAD)..HEAD
• Mid-chain → git log $LAST_REVIEW_SHA..HEAD ($LAST_REVIEW_SHA = most-recent trail record via list-review-trail-records.sh | tail -1 whose sha_range head passes git merge-base --is-ancestor <sha> HEAD; iterate oldest to newest; fall back to session-start SHA if none passes).

Dispatch: invoke coordinator:review-code Branch A.2 with the resolved diff scope.

Spec cross-reference (loop closure) — include in dispatch brief when a spec exists: When work is governed by a spec/plan/stub, name the spec path in the code-reviewer dispatch brief and instruct it to apply the Spec completion lens (per agents/code-reviewer.md § Spec completion lens). Apply on row 3/4/5/6 sessions; omit on row 1/2. If multiple specs apply, name all of them; the reviewer treats the union as the completion oracle. When partitioning the diff (§ Partitioning large surfaces), name each reviewer's spec slice explicitly.

Negative-spec: if no spec governs this session, omit the spec section from the brief — do not invent one. No spec named ⇒ reviewer skips the lens entirely.

Findings disposition — fix everything, including nitpicks. Every severity (P0/P1/P2/nitpick/observation/'consider') folds in via coordinator:review-integrator before the marker-trail write. "Recorded below blocking threshold" in the wrap-up is the tell that this rule was skipped — re-open and fold. Only legitimate skip: real tradeoff → escalate to PM (coordinator CLAUDE.md § Reviewer findings — apply, don't ratify).

The trail's --verdict field records the reviewer's pre-fix verdict (ok/warn/blocked), not what shipped — downstream load-shedding consumes the verdict; the trail is not a fix-completion log.

Marker write: after review integration completes, invoke:

~/.claude/plugins/coordinator/bin/coordinator-write-review-trail.sh \
  --sha-range <A..B> --reviewer <code-reviewer|patrik|code-reviewer+patrik|waived|ubt-compile> \
  --scope <chain|session> --verdict <ok|warn|blocked|waived|pending> --diff-loc <N>

Negative-spec:

• Trivial sessions (Row 1, 2 of the table): skip the review entirely. No trail record written.
• PM-waived sessions: log waiver to trail with --reviewer waived --verdict waived. Greppable as verdict=waived.

Staging discipline: files edited by coordinator:review-integrator must be staged via explicit path in Step 3 — not git add -A.

UBT pending-marker (UE plugin work only): If bin/check-ubt-build-fresh.sh exists in the cwd, invoke it in --mode pending. Captures the build verdict as a deferred record; resolution happens at /workday-complete Step 0c. This step is a no-op for non-UE repos (script absent) — the [ -x bin/<name>.sh ] pattern is the canonical convention for conditional UE-specific steps in coordinator skill bodies; future UE conditionals (clippy, etc.) follow this shape.

[ -x bin/check-ubt-build-fresh.sh ] && \
  bin/check-ubt-build-fresh.sh --since "$(git merge-base origin/main HEAD 2>/dev/null || git rev-parse HEAD~1)" --mode pending

Step 2.95: Aftermath Lens-Checkers (big-workstream sessions)

Fires on big workstreams — same trigger as Step 2.9 rows 3/4/5/6. Skip silently on row-1/row-2 trivial sessions.

Step 2.9 is the line-level lens; these are cross-cutting. Quick self-check (not a reviewer dispatch unless depth needed):

1. Install-surface completeness — wrote state outside source (machine-local/, installer scripts, sentinels, env)? Confirm clean-install reproduces it. → docs/wiki/install-surface-completeness.md.
2. Contact-point / convention drift — added a convention, tripwire, hook, or agent-facing surface? Every contact-point updated in this workstream. → CLAUDE.md § Adding a Convention.
3. Doc + wiki alignment — big changes age docs/README.md, wiki, atlas; repoint stale refs (don't duplicate).
4. Lessons + queue capture — pattern worth promoting? Per Step 1/1.2; universals to central queue.
5. Security / secret-exposure — touched auth, secrets, external APIs, new deps? Route to security-audit-worker / dep-cve-auditor (CLAUDE.md § Reviewer-Routed Workers; docs/wiki/reviewer-routed-workers.md) — don't self-assess.

Output: one-line-per-lens checklist in Step 4 summary (clear / <finding + disposition>). Tradeoff-free corrections fold in; real tradeoffs surface to PM. "All clear" must be affirmative, not silent.

Step 3: Commit + Verify Remote

Step 3.0: Pre-terminate dirty-tree gate

Pre-terminate dirty-tree gate (fail loud on unattributable files). Before the workstream-complete commit, run git status --porcelain and classify every dirty path. EOL phantoms are benign — never case (c): a file where git diff --quiet -- <path> exits 0 is a Git-for-Windows stat-staleness artifact (docs/wiki/concurrent-em-hazards.md § H23) — leave it untouched; swept by coordinator-renormalize-index at session start. Classify each remaining path:

• (a) This session authored it → it belongs in this terminator's scoped commit (handled by the existing scope/commit step).
• (b) A known concurrent session owns it → leave it alone. "Known" means you can name the workstream/session — a sibling scope: block, an active handoff, or a consumed_by: field in handoff frontmatter naming another session's id.
• (c) Unattributable — a dirty file you did NOT author AND cannot tie to a named concurrent owner (the classic case: an abandoned partial revert or orphaned edit from a crashed session). Do NOT silently leave these — they wedge the next session opener. Fail loud and pick exactly one disposition, in this order of preference:
  1. Commit with provenance if the change is coherent and you can attribute it: git add -- <path> && git commit -m "chore: adopt orphaned WT change <path> — unattributed at workstream-complete".
  2. Stash-with-provenance if it is incoherent or risky to commit: git stash push -u -m "orphaned-WT <YYYY-MM-DD> workstream-complete: <path> — left by unknown session" -- <path>. Name the stash so the next session can find and adjudicate it (per CLAUDE.md "Probe edits in git stash push -u / pop").
  3. Explicit "leave it owned by X" only when you can now name the owner — record a one-line note (in the handoff body / session summary) stating which session/workstream owns it, converting it from case (c) to case (b).

The forbidden outcome is terminating with case-(c) files still dirty and unnamed. Orphan .tmp.<pid>.<nanos> files = Edit-tool atomic-write crash (CLAUDE.md § Verifying Executor Output) — diff against target before deleting; do not stash blind.

1. Stage only paths this session touched — never git add -A. Name each path explicitly: git add <path1> <path2> .... Typical set: state/lessons.md, docs/plans/<feature>.md (if Step 2.4 ran), archive/completed/YYYY-MM/<entry>.md, docs/project-tracker.md, action-items, docs/README.md, state/handoff-tracker.md (if Step 2.75 ran). Unfamiliar dirty files → Step 3.0 gate first; "leave alone" is only correct for case (b) named-owner files.
2. Commit with a lightweight message: "workstream-complete quick-save". (The post-commit hook will auto-push on work/feature branches.)
3. If nothing to commit, check for unpushed commits: git log "origin/$(~/.claude/plugins/coordinator/bin/coordinator-current-branch)..HEAD" 2>/dev/null
4. Verify remote is synced: confirm no unpushed commits remain. If auto-push failed, push explicitly and warn the PM.
5. If on main (shouldn't happen, but safety): push explicitly — git push origin main
6. If push fails (auth, network, conflicts), warn the PM explicitly — this is a critical failure

Step 3.5: Archive Session Claim

Archive this session's claim directory — required at both /workstream-complete and /handoff to avoid forcing the next concurrent EM into a 24h wait.

Run:

sid="${CLAUDE_CODE_SESSION_ID:-$(cat "$(git rev-parse --show-toplevel)/.git/coordinator-sessions/.current-session-id" 2>/dev/null)}" && \
  source ~/.claude/plugins/coordinator/lib/coordinator-session.sh 2>/dev/null && \
  cs_archive "$sid" 2>/dev/null || true

Idempotent. Failures are non-fatal (24h reaper is the safety net). Skip if session id can't be resolved or lib is unavailable. Prefer $CLAUDE_CODE_SESSION_ID; .current-session-id is last-writer-wins fallback only.

Step 3.8: Acceptance-Oracle Gate (AUTHORITATIVE)

<!-- spec-backlink: docs/wiki/acceptance-oracle.md § Where checked — gate seam relocated from /merge-to-main 2026-06-02 -->

If this session executed an oracle-bearing plan (one that went through coordinator:review and carries a bindable ## Acceptance Criteria table with gate-bound rows), this is the authoritative gate. One workstream = one plan = one AC table in frame; the oracle is load-bearing here, not at merge.

Plan-path discovery (try in order):

1. Frontmatter plan: field on the workstream's plan document.
2. Explicit --plan <path> flag in $ARGUMENTS.
3. If neither yields a path AND the session shape is plan-execution → skip-with-offer: "No plan path found — acceptance oracle can be validated manually with bash check-acceptance-oracle.sh <plan-path>." Continue to Step 4.

If COORDINATOR_OVERRIDE_ACCEPTANCE_GATE=1 is set: Skip the gate. Log: "Acceptance-oracle gate bypassed via COORDINATOR_OVERRIDE_ACCEPTANCE_GATE=1 — exceptional use only." Continue to Step 4.

If plan path resolved AND plan contains a bindable ## Acceptance Criteria table:

bash check-acceptance-oracle.sh <plan-path>

• Exit 0 (all gate-bound rows green or cited-resolved): Log the verdict. "Acceptance oracle: all gate-bound tests pass — workstream may complete." Continue to Step 4.
• Non-zero exit (any gate-bound row red or unresolved): Hard-block. Print: "Workstream-complete blocked: acceptance oracle has red/unresolved gate-bound tests." + the script verdict. Remediation: fix tests and re-run, OR mark deviations via ## Deviations + Status → shipped-differently with a cited: row, OR set COORDINATOR_OVERRIDE_ACCEPTANCE_GATE=1 (exceptional — cited: rows are the routine accommodation). Stop. Downstream /merge-to-main trusts this seam.

If plan path resolved but no bindable ## Acceptance Criteria table (old-form plan): skip-with-offer: "Plan found but no bindable acceptance-criteria table — oracle gate skipped. Consider upgrading (docs/wiki/writing-plans.md § Acceptance Oracle)."

If no oracle-bearing plan was involved (doctrine/sweep/memo-action without a reviewed plan): skip silently. Daily-rollups and non-plan workstreams have nothing to gate on.

Why authoritative here, not at merge: /merge-to-main aggregates workstreams + doctrine edits + sweeps — no single AC table governs that union. The oracle is load-bearing where one plan is in frame.

Step 4: Final Summary

Present a brief end-of-session summary:

## Session Complete

**Work done:** [1-2 sentence summary]
**Lessons captured:** [N new / none]
**Work archived:** [N items written to archive/completed/YYYY-MM/<filename>.md / none needed / project not using unified tracking]
**Docs updated:** [list of updated files]
**Orientation refreshed:** [orientation cache patched / tracker updated / action items checked off / nothing to update / no orientation docs exist]
**Pushed to remote:** [yes — branch name / no — reason]

Flag to PM: Explicitly note the push so they can verify nothing breaks for other consumers.

If $ARGUMENTS is provided, use it as context for what was accomplished this session.