gauntlet 2026-05-02: smoke chain + per-component scrum + parity probe

Production-readiness gauntlet exploiting the dual Rust/Go implementation as a measurement instrument. ## Phase 1 — Full smoke chain 21/21 PASS in ~60s. Substrate intact across the full service surface. ## Phase 2 — Per-component scrum (token-volume fix) Prior wave (165KB diff): Kimi 62 tokens out, Qwen 297 → no useful analysis. This wave splits today's commits into 4 focused bundles (36-71KB each): c1 validatord (46KB) → 0 convergent / 11 distinct c2 vectord substrate (36KB) → 0 convergent / 10 distinct c3 materializer (71KB) → 0 convergent / 6 distinct (Opus emitted a BLOCK then self-retracted in same response) c4 replay (45KB) → 0 convergent / 10 distinct Reviewer engagement vs prior wave: Kimi went 62 → ~250 tokens out once bundles dropped below 60KB. scripts/scrum_review.sh hardening: * Diff-size guard (warn >60KB, hard-fail >100KB, SCRUM_FORCE_OVERSIZE=1 override) * Tightened prompt — file path must appear EXACTLY as in diff so post-processor can grep WHERE: lines reliably * Auto-tally step dedupes by (reviewer, location); convergence counts distinct lineages (closes the prior `opus+opus+opus` false-convergence bug) ## Phase 3 — Cross-runtime validator parity probe (the headline finding) scripts/cutover/parity/validator_parity.sh sends 6 identical /v1/validate cases to Rust :3100 AND Go :4110, compares status+body. Result: **6/6 status codes match · 5/6 body shapes diverge.** Rust returns serde-tagged enum: {"Schema":{"field":"x","reason":"y"}} Go returns flat exported-fields: {"Kind":"schema","Field":"x","Reason":"y"} Both round-trip inside their own runtime; a caller swapping one for the other would break parsing silently. Captured as new _open_ row in docs/ARCHITECTURE_COMPARISON.md decisions tracker. This is the "use the dual-implementation as a measurement instrument" return — single-repo scrums can't catch this class of cross-runtime drift. ## Phase 4 — Production assessment ship-with-known-gap. Validator wire-format gap is documented, not regressed. ~50 LOC future fix on Go side (custom MarshalJSON on ValidationError to match Rust's serde shape). Persistent stack config (/tmp/lakehouse-persistent.toml) gains validatord on :3221 + persistent-validatord binary so operators bringing up the persistent stack get the new daemon automatically. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-02 04:05:18 -05:00
parent f9e72412c1
commit e8cf113af8
28 changed files with 7195 additions and 5 deletions
@@ -0,0 +1,966 @@
+commit 89ca72d4718fcb20ba9dcc03110e090890a0736e
+Author: root <root@island37.com>
+Date:   Sat May 2 03:31:02 2026 -0500
+
+    materializer + replay ports + vectord substrate fix verified at scale
+    
+    Two threads landing together — the doc edits interleave so they ship
+    in a single commit.
+    
+    1. **vectord substrate fix verified at original scale** (closes the
+       2026-05-01 thread). Re-ran multitier 5min @ conc=50: 132,211
+       scenarios at 438/sec, 6/6 classes at 0% failure (was 4/6 pre-fix).
+       Throughput dropped 1,115 → 438/sec because previously-broken
+       scenarios now do real HNSW Add work — honest cost of correctness.
+       The fix (i.vectors side-store + safeGraphAdd recover wrappers +
+       smallIndexRebuildThreshold=32 + saveTask coalescing) holds at the
+       footprint that originally surfaced the bug.
+    
+    2. **Materializer port** — internal/materializer + cmd/materializer +
+       scripts/materializer_smoke.sh. Ports scripts/distillation/transforms.ts
+       (12 transforms) + build_evidence_index.ts (idempotency, day-partition,
+       receipt). On-wire JSON shape matches TS so Bun and Go runs are
+       interchangeable. 14 tests green.
+    
+    3. **Replay port** — internal/replay + cmd/replay +
+       scripts/replay_smoke.sh. Ports scripts/distillation/replay.ts
+       (retrieve → bundle → /v1/chat → validate → log). Closes audit-FULL
+       phase 7 live invocation on the Go side. Both runtimes append to the
+       same data/_kb/replay_runs.jsonl (schema=replay_run.v1). 14 tests green.
+    
+    Side effect on internal/distillation/types.go: EvidenceRecord gained
+    prompt_tokens, completion_tokens, and metadata fields to mirror the TS
+    shape the materializer transforms produce.
+    
+    STATE_OF_PLAY refreshed to 2026-05-02; ARCHITECTURE_COMPARISON decisions
+    tracker moves the materializer + replay items from _open_ to DONE and
+    adds the substrate-fix scale verification row.
+    
+    Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
+
+diff --git a/cmd/vectord/main.go b/cmd/vectord/main.go
+index 9bab5e3..c76b9aa 100644
+--- a/cmd/vectord/main.go
+++ b/cmd/vectord/main.go
+@@ -17,6 +17,7 @@ import (
+ 	"os"
+ 	"strconv"
+ 	"strings"
+	"sync"
+ 	"time"
+ 
+ 	"github.com/go-chi/chi/v5"
+@@ -71,6 +72,73 @@ func main() {
+ type handlers struct {
+ 	reg     *vectord.Registry
+ 	persist *vectord.Persistor // nil when persistence is disabled
+
+	// saversMu guards lazy initialization of per-index save tasks.
+	// Each task coalesces synchronous Save calls into single-flight
+	// async saves so high-write-rate indexes (playbook_memory under
+	// multitier_100k load) don't pay one MinIO PUT per Add. See the
+	// saveTask docstring for the coalescing semantics.
+	saversMu sync.Mutex
+	savers   map[string]*saveTask
+}
+
+// saveTask coalesces saves for one index into a single-flight async
+// goroutine. While a save is in-flight, additional triggers mark
+// "pending" — the in-flight goroutine reruns the save after it
+// finishes, collapsing N concurrent triggers into at most 2 saves
+// (the current in-flight + one catch-up).
+//
+// Why: pre-2026-05-01 each successful Add called Persistor.Save
+// synchronously inside the request handler. For playbook_memory at
+// 1900-entry / 768-d, Encode + MinIO PUT cost 100-300ms. With 50
+// concurrent writers, end-to-end Add latency hit 2-2.5s purely from
+// save serialization (Save takes the index RLock for Encode, which
+// blocks new Adds taking the Lock).
+//
+// Trade-off: RPO. Add now returns OK before the save completes, so
+// a crash can lose up to ~1 save's worth of data. Acceptable for
+// the playbook-memory shape (learning loop — lost trace re-recorded
+// on next run) and consistent with ADR-005's fail-open posture.
+type saveTask struct {
+	mu       sync.Mutex
+	inflight bool
+	pending  bool
+}
+
+// trigger schedules a save. If a save is already in-flight, marks
+// pending and returns. If none in-flight, starts a goroutine that
+// runs save and any queued pending saves.
+//
+// save is the actual save operation (parameterized for testability).
+// Errors are logged via slog and not returned — same fail-open
+// posture as the prior synchronous saveAfter.
+func (s *saveTask) trigger(save func() error) {
+	s.mu.Lock()
+	if s.inflight {
+		s.pending = true
+		s.mu.Unlock()
+		return
+	}
+	s.inflight = true
+	s.mu.Unlock()
+
+	go func() {
+		for {
+			if err := save(); err != nil {
+				slog.Warn("persist save", "err", err)
+			}
+			s.mu.Lock()
+			if !s.pending {
+				s.inflight = false
+				s.mu.Unlock()
+				return
+			}
+			s.pending = false
+			s.mu.Unlock()
+			// Loop: re-run save to capture changes that arrived
+			// while we were saving.
+		}
+	}()
+ }
+ 
+ // rehydrate enumerates persisted indexes and loads each into the
+@@ -103,19 +171,38 @@ func (h *handlers) rehydrate(ctx context.Context) (int, error) {
+ 	return loaded, nil
+ }
+ 
+-// saveAfter is the post-write persistence hook. Logs-not-fatal:
+-// in-memory state is the source of truth in flight; a failed save
+-// gets re-attempted on the next mutation, and the operator log
+-// shows the storaged outage.
+// saveAfter triggers a coalesced async persistence for the index.
+// In-memory state is the source of truth in flight; a failed save
+// re-runs on the next mutation, and the operator log shows the
+// storaged outage.
+//
+// Coalescing semantics (added 2026-05-01 after multitier_100k
+// follow-up): rapid concurrent writes collapse into at most two
+// MinIO PUTs per index (current + one catch-up), instead of one
+// per Add. See the saveTask docstring.
+ func (h *handlers) saveAfter(idx *vectord.Index) {
+ 	if h.persist == nil {
+ 		return
+ 	}
+-	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+-	defer cancel()
+-	if err := h.persist.Save(ctx, idx); err != nil {
+-		slog.Warn("persist save", "name", idx.Params().Name, "err", err)
+	name := idx.Params().Name
+	h.saversMu.Lock()
+	if h.savers == nil {
+		h.savers = make(map[string]*saveTask)
+	}
+	s, ok := h.savers[name]
+	if !ok {
+		s = &saveTask{}
+		h.savers[name] = s
+ 	}
+	h.saversMu.Unlock()
+	s.trigger(func() error {
+		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+		defer cancel()
+		if err := h.persist.Save(ctx, idx); err != nil {
+			return err
+		}
+		return nil
+	})
+ }
+ 
+ // deleteAfter mirrors saveAfter for the Delete path.
+diff --git a/cmd/vectord/main_test.go b/cmd/vectord/main_test.go
+index 045924d..fa13ed8 100644
+--- a/cmd/vectord/main_test.go
+++ b/cmd/vectord/main_test.go
+@@ -3,11 +3,15 @@ package main
+ import (
+ 	"bytes"
+ 	"encoding/json"
+	"errors"
+ 	"net/http"
+ 	"net/http/httptest"
+ 	"strconv"
+ 	"strings"
+	"sync"
+	"sync/atomic"
+ 	"testing"
+	"time"
+ 
+ 	"github.com/go-chi/chi/v5"
+ 
+@@ -417,3 +421,105 @@ func TestSearchK_DefaultsAndMax(t *testing.T) {
+ 		t.Errorf("maxK=%d unreasonably large", maxK)
+ 	}
+ }
+
+// TestSaveTask_Coalesces locks the multitier_100k follow-up: a
+// burst of triggers must collapse into at most 2 actual saves
+// (the in-flight one + one catch-up). Without coalescing, every
+// trigger would yield a save and concurrent writers would
+// serialize on the index RLock during Encode (the original
+// 1-2.5s tail-latency cause).
+func TestSaveTask_Coalesces(t *testing.T) {
+	var (
+		s        saveTask
+		saveCnt  atomic.Int32
+		started  = make(chan struct{}, 1)
+		release  = make(chan struct{})
+	)
+	save := func() error {
+		// First save blocks until released so we can pile up
+		// triggers behind it. Subsequent saves return fast so the
+		// catch-up logic completes promptly.
+		n := saveCnt.Add(1)
+		if n == 1 {
+			started <- struct{}{}
+			<-release
+		}
+		return nil
+	}
+	// Trigger first save and wait for it to enter the blocked region.
+	s.trigger(save)
+	<-started
+	// Pile up triggers while the first is blocked. None of these
+	// should start their own goroutines — they should mark "pending".
+	for i := 0; i < 50; i++ {
+		s.trigger(save)
+	}
+	// Release the first save. The trigger logic should run ONE
+	// catch-up save for all 50 piled-up triggers, then return.
+	close(release)
+	// Wait for the goroutine to drain.
+	deadline := time.Now().Add(2 * time.Second)
+	for time.Now().Before(deadline) {
+		s.mu.Lock()
+		idle := !s.inflight && !s.pending
+		s.mu.Unlock()
+		if idle {
+			break
+		}
+		time.Sleep(5 * time.Millisecond)
+	}
+	got := saveCnt.Load()
+	if got != 2 {
+		t.Errorf("save count = %d, want 2 (one in-flight + one catch-up)", got)
+	}
+}
+
+// TestSaveTask_RunsOnce — single trigger fires exactly one save.
+func TestSaveTask_RunsOnce(t *testing.T) {
+	var s saveTask
+	var n atomic.Int32
+	done := make(chan struct{})
+	s.trigger(func() error {
+		n.Add(1)
+		close(done)
+		return nil
+	})
+	select {
+	case <-done:
+	case <-time.After(2 * time.Second):
+		t.Fatal("trigger goroutine never ran")
+	}
+	// Wait briefly for the goroutine to mark inflight=false.
+	time.Sleep(20 * time.Millisecond)
+	if got := n.Load(); got != 1 {
+		t.Errorf("save count = %d, want 1", got)
+	}
+}
+
+// TestSaveTask_LogsSaveError — a save error doesn't break the
+// coalescing state machine; subsequent triggers still work.
+func TestSaveTask_LogsSaveError(t *testing.T) {
+	var s saveTask
+	var n atomic.Int32
+	wantErr := errors.New("boom")
+	var wg sync.WaitGroup
+	wg.Add(1)
+	s.trigger(func() error {
+		defer wg.Done()
+		n.Add(1)
+		return wantErr
+	})
+	wg.Wait()
+	// State must reset so the next trigger fires another save.
+	time.Sleep(20 * time.Millisecond)
+	wg.Add(1)
+	s.trigger(func() error {
+		defer wg.Done()
+		n.Add(1)
+		return nil
+	})
+	wg.Wait()
+	if got := n.Load(); got != 2 {
+		t.Errorf("save count = %d, want 2 (failure must not stall the task)", got)
+	}
+}
+diff --git a/internal/vectord/index.go b/internal/vectord/index.go
+index 20e1710..95d4495 100644
+--- a/internal/vectord/index.go
+++ b/internal/vectord/index.go
+@@ -33,6 +33,23 @@ const (
+ 	DefaultEfSearch = 20
+ )
+ 
+// smallIndexRebuildThreshold guards against coder/hnsw v0.6.1's
+// degenerate-state nil-deref (graph.go:95 layerNode.search) which
+// fires when the graph transitions through low-len states with a
+// stale entry pointer. Below this threshold, Add and BatchAdd
+// rebuild the entire graph from scratch — fresh graph + one
+// variadic Add never exercises the buggy incremental path.
+//
+// Why 32: HNSW's value is sub-linear search at large N; at N<32 a
+// rebuild's O(n) cost (snapshot ids + bulk Add) is negligible
+// (~µs at 768-d). The boundary is intentionally above the small
+// playbook-corpus regime (where multitier_100k surfaced the bug)
+// but well below realistic working-set indexes.
+//
+// The recover() guard in BatchAdd remains as belt-and-suspenders
+// for any incremental-path edge cases past the threshold.
+const smallIndexRebuildThreshold = 32
+
+ // IndexParams describes one vector index. Once an Index is built,
+ // these are fixed — changing M / dimension / distance requires a
+ // rebuild.
+@@ -55,21 +72,30 @@ type Result struct {
+ 	Metadata json.RawMessage `json:"metadata,omitempty"`
+ }
+ 
+-// Index wraps a coder/hnsw graph plus a side map of opaque JSON
+-// metadata per ID. Concurrency: read-heavy via Search (read-lock);
+-// Add and Delete take the write lock.
+// Index wraps a coder/hnsw graph plus side maps of opaque JSON
+// metadata and raw vectors per ID. Concurrency: read-heavy via
+// Search (read-lock); Add and Delete take the write lock.
+//
+// Why we keep vectors in a side map (i.vectors) in addition to the
+// graph: coder/hnsw v0.6.1 has a known bug where the graph
+// transitions through degenerate states after Delete cycles, and
+// later operations (Add / Lookup) can panic with nil-deref. The
+// side map is independent of graph state, so the rebuild path can
+// always reconstruct a clean graph even if the current one is
+// corrupted. Memory cost is ~2x for vectors (also held in graph),
+// which is acceptable for the safety it buys. Verified necessary
+// 2026-05-01 multitier_100k where the bug fired at len=40.
+ type Index struct {
+ 	params IndexParams
+ 	g      *hnsw.Graph[string]
+ 	meta   map[string]json.RawMessage
+-	// ids is the canonical ID set (a value-less map used as a set).
+-	// Maintained alongside i.g and i.meta in Add/Delete/resetGraph
+-	// so IDs() can enumerate without depending on the meta map's
+-	// sparse-on-nil-meta semantics. Underpins OPEN #1's merge
+-	// endpoint — necessary because two-tier callers
+-	// (multi_coord_stress et al.) sometimes Add with nil meta.
+-	ids map[string]struct{}
+-	mu  sync.RWMutex
+	// vectors is the panic-safe source of truth — every successful
+	// Add stores the vector here, every Delete removes it, and
+	// rebuildGraphLocked reads from this map (not i.g.Lookup) so
+	// it tolerates a corrupted graph. Map keys are also the
+	// canonical ID set (replaces the prior i.ids map).
+	vectors map[string][]float32
+	mu      sync.RWMutex
+ }
+ 
+ // Errors surfaced to HTTP handlers. Sentinel-based so the wire
+@@ -110,10 +136,10 @@ func NewIndex(p IndexParams) (*Index, error) {
+ 	// is a G2 concern when we have real tuning data.
+ 
+ 	return &Index{
+-		params: p,
+-		g:      g,
+-		meta:   make(map[string]json.RawMessage),
+-		ids:    make(map[string]struct{}),
+		params:  p,
+		g:       g,
+		meta:    make(map[string]json.RawMessage),
+		vectors: make(map[string][]float32),
+ 	}, nil
+ }
+ 
+@@ -133,10 +159,14 @@ func distanceFn(name string) (hnsw.DistanceFunc, error) {
+ func (i *Index) Params() IndexParams { return i.params }
+ 
+ // Len returns the number of vectors currently in the index.
+//
+// Reads from i.vectors (the panic-safe source of truth) rather
+// than i.g.Len() — the latter can drift past Len during a corrupted
+// graph state. i.vectors only changes on successful Add/Delete.
+ func (i *Index) Len() int {
+ 	i.mu.RLock()
+ 	defer i.mu.RUnlock()
+-	return i.g.Len()
+	return len(i.vectors)
+ }
+ 
+ // IDs returns a snapshot of every ID currently stored in the index.
+@@ -145,16 +175,15 @@ func (i *Index) Len() int {
+ // (OPEN #1: periodic fresh→main index merge — drains the fresh
+ // corpus into the main one when it crosses the operational ceiling).
+ //
+-// Source of truth: the i.ids tracker, NOT the meta map. The meta
+-// map intentionally stays sparse (only items with explicit
+-// metadata appear there, per the K-B1 nil-vs-{} distinction). Using
+-// meta as the ID set would silently miss items added with nil
+-// metadata.
+// Source of truth: the i.vectors keyset. The meta map stays sparse
+// (only items with explicit metadata appear there, per the K-B1
+// nil-vs-{} distinction); using meta as the ID set would silently
+// miss items added with nil metadata.
+ func (i *Index) IDs() []string {
+ 	i.mu.RLock()
+ 	defer i.mu.RUnlock()
+-	out := make([]string, 0, len(i.ids))
+-	for id := range i.ids {
+	out := make([]string, 0, len(i.vectors))
+	for id := range i.vectors {
+ 		out = append(out, id)
+ 	}
+ 	return out
+@@ -191,23 +220,38 @@ func (i *Index) Add(id string, vec []float32, meta json.RawMessage) error {
+ 	}
+ 	i.mu.Lock()
+ 	defer i.mu.Unlock()
+-	// coder/hnsw has two sharp edges on re-add:
+-	//   1. Add of an existing key panics with "node not added"
+-	//      (length-invariant fires because internal delete+re-add
+-	//      doesn't change Len). Pre-Delete fixes this for n>1.
+-	//   2. Delete of the LAST node leaves layers[0] non-empty but
+-	//      entryless; the next Add SIGSEGVs in Dims() because
+-	//      entry().Value is nil. We rebuild the graph in that case.
+-	_, exists := i.g.Lookup(id)
+-	if exists {
+-		if i.g.Len() == 1 {
+-			i.resetGraphLocked()
+-		} else {
+-			i.g.Delete(id)
+	// Re-add: drop existing graph entry AND side-store entry before
+	// the new Add. Without removing from i.vectors, the rebuild path
+	// below would see both old and new entries and double-add.
+	// safeGraphDelete tolerates a corrupted graph; i.vectors is
+	// authoritative regardless.
+	if _, exists := i.vectors[id]; exists {
+		_ = safeGraphDelete(i.g, id)
+		delete(i.vectors, id)
+	}
+	newNode := hnsw.MakeNode(id, vec)
+	postLen := len(i.vectors) + 1
+	addOK := false
+	if postLen <= smallIndexRebuildThreshold {
+		i.rebuildGraphLocked([]hnsw.Node[string]{newNode})
+		addOK = true
+	} else {
+		// Warm path: try incremental Add. If the graph is in a
+		// degenerate state from a prior Delete cycle, this panics;
+		// we recover and rebuild from the panic-safe i.vectors map.
+		addOK = safeGraphAdd(i.g, newNode)
+		if !addOK {
+			i.rebuildGraphLocked([]hnsw.Node[string]{newNode})
+			addOK = true
+ 		}
+ 	}
+-	i.g.Add(hnsw.MakeNode(id, vec))
+-	i.ids[id] = struct{}{}
+	if !addOK {
+		return errors.New("vectord: hnsw add failed even after rebuild — should never happen")
+	}
+	// Commit to the side stores after the graph mutation succeeded.
+	out := make([]float32, len(vec))
+	copy(out, vec)
+	i.vectors[id] = out
+ 	if meta != nil {
+ 		// Per scrum K-B1 (Kimi): only OVERWRITE on explicit non-nil.
+ 		// nil = "leave existing meta alone" (upsert). To clear, the
+@@ -217,17 +261,59 @@ func (i *Index) Add(id string, vec []float32, meta json.RawMessage) error {
+ 	return nil
+ }
+ 
+-// resetGraphLocked recreates the underlying coder/hnsw Graph with
+-// the same params. Caller MUST hold i.mu (write-lock). Used to
+-// dodge the library's "delete the last node, then segfault on
+-// next Add" bug — see Add for details. Metadata map is preserved
+-// because the only entry it could affect is the one being
+-// re-added, which Add overwrites.
+-func (i *Index) resetGraphLocked() {
+// safeGraphAdd wraps coder/hnsw's variadic Graph.Add with a
+// recover() so v0.6.1's degenerate-state nil-deref returns false
+// instead of crashing the caller. Caller is expected to fall back
+// to rebuildGraphLocked on false.
+func safeGraphAdd(g *hnsw.Graph[string], nodes ...hnsw.Node[string]) (ok bool) {
+	defer func() {
+		if r := recover(); r != nil {
+			ok = false
+		}
+	}()
+	g.Add(nodes...)
+	return true
+}
+
+// safeGraphDelete wraps Graph.Delete with recover for the same
+// reason — Delete can also touch corrupted layer state.
+func safeGraphDelete(g *hnsw.Graph[string], id string) (ok bool) {
+	defer func() {
+		if r := recover(); r != nil {
+			ok = false
+		}
+	}()
+	return g.Delete(id)
+}
+
+// rebuildGraphLocked replaces i.g with a fresh graph containing
+// the current items (snapshotted from the panic-safe i.vectors
+// map) plus the supplied extras, in one bulk Add into a freshly-
+// created graph. Caller MUST hold the write lock.
+//
+// Independence from i.g state is the load-bearing property — even
+// if i.g is corrupted from a prior coder/hnsw v0.6.1 panic, this
+// rebuild produces a clean graph because i.vectors is maintained
+// only on successful Add/Delete.
+//
+// Caller MUST ensure that any extra IDs already present in
+// i.vectors have been removed first (otherwise the bulk Add will
+// see duplicate IDs and panic).
+func (i *Index) rebuildGraphLocked(extras []hnsw.Node[string]) {
+ 	g := hnsw.NewGraph[string]()
+ 	g.M = i.params.M
+ 	g.EfSearch = i.params.EfSearch
+ 	g.Distance = i.g.Distance
+
+	nodes := make([]hnsw.Node[string], 0, len(i.vectors)+len(extras))
+	for id, vec := range i.vectors {
+		nodes = append(nodes, hnsw.MakeNode(id, vec))
+	}
+	nodes = append(nodes, extras...)
+
+	if len(nodes) > 0 {
+		g.Add(nodes...)
+	}
+ 	i.g = g
+ }
+ 
+@@ -296,17 +382,15 @@ func (i *Index) BatchAdd(items []BatchItem) error {
+ 	i.mu.Lock()
+ 	defer i.mu.Unlock()
+ 
+-	// Pre-pass: drop any existing IDs so coder/hnsw's variadic Add
+-	// never sees a re-add. Same library-quirk handling as single
+-	// Add — Len()==1 needs a full graph reset because Delete of the
+-	// last node leaves layers[0] entryless.
+	// Pre-pass: drop any existing IDs from BOTH the graph and the
+	// side-store map so the rebuild snapshot doesn't double-add and
+	// the warm path's variadic Add never sees a re-add. Graph Delete
+	// is wrapped in safeGraphDelete because corrupted graphs can also
+	// panic on Delete; the side store remains authoritative.
+ 	for _, it := range items {
+-		if _, exists := i.g.Lookup(it.ID); exists {
+-			if i.g.Len() == 1 {
+-				i.resetGraphLocked()
+-			} else {
+-				i.g.Delete(it.ID)
+-			}
+		if _, exists := i.vectors[it.ID]; exists {
+			_ = safeGraphDelete(i.g, it.ID)
+			delete(i.vectors, it.ID)
+ 		}
+ 	}
+ 
+@@ -314,27 +398,26 @@ func (i *Index) BatchAdd(items []BatchItem) error {
+ 	for j, it := range items {
+ 		nodes[j] = hnsw.MakeNode(it.ID, it.Vector)
+ 	}
+-	// coder/hnsw v0.6.1 has a known nil-deref in layerNode.search at
+-	// graph.go:95 when the graph transitions through degenerate
+-	// states (len=0/1 with stale entry from a prior Delete cycle).
+-	// Wrap with recover so a panic becomes an error rather than
+-	// killing the request handler. Surfaced under sustained
+-	// playbook_record load (multitier test 2026-05-01); operator
+-	// recovery is `DELETE /vectors/index/<name>` then re-record.
+-	if addErr := func() (err error) {
+-		defer func() {
+-			if r := recover(); r != nil {
+-				err = fmt.Errorf("hnsw add panic (coder/hnsw v0.6.1 small-index bug — DELETE the index to recover): %v", r)
+-			}
+-		}()
+-		i.g.Add(nodes...)
+-		return nil
+-	}(); addErr != nil {
+-		return addErr
+
+	// Below threshold: rebuild from scratch unconditionally — fresh
+	// graph + one bulk Add never exercises v0.6.1's degenerate-state
+	// path. At/above threshold: try warm incremental Add, fall back
+	// to rebuild on panic. The rebuild always succeeds because
+	// i.vectors is independent of graph state.
+	postLen := len(i.vectors) + len(nodes)
+	if postLen <= smallIndexRebuildThreshold {
+		i.rebuildGraphLocked(nodes)
+	} else {
+		if !safeGraphAdd(i.g, nodes...) {
+			i.rebuildGraphLocked(nodes)
+		}
+ 	}
+ 
+	// Commit to side stores after the graph is in good shape.
+ 	for _, it := range items {
+-		i.ids[it.ID] = struct{}{}
+		out := make([]float32, len(it.Vector))
+		copy(out, it.Vector)
+		i.vectors[it.ID] = out
+ 		if it.Metadata != nil {
+ 			i.meta[it.ID] = it.Metadata
+ 		}
+@@ -374,12 +457,22 @@ func dedupBatchLastWins(items []BatchItem) []BatchItem {
+ }
+ 
+ // Delete removes id from the index. Returns true if present.
+//
+// The side store i.vectors is the authority on presence; the graph
+// Delete is best-effort (can panic on corrupted state, recovered
+// via safeGraphDelete). The side store always reflects the
+// post-Delete truth so the next rebuild produces a clean graph.
+ func (i *Index) Delete(id string) bool {
+ 	i.mu.Lock()
+ 	defer i.mu.Unlock()
+	_, present := i.vectors[id]
+	if !present {
+		return false
+	}
+ 	delete(i.meta, id)
+-	delete(i.ids, id)
+-	return i.g.Delete(id)
+	delete(i.vectors, id)
+	_ = safeGraphDelete(i.g, id)
+	return true
+ }
+ 
+ // Search returns the k nearest neighbors of query, sorted
+@@ -456,9 +549,9 @@ func (i *Index) Encode(envelopeW, graphW io.Writer) error {
+ 	defer i.mu.RUnlock()
+ 
+ 	// v2: serialize the canonical ID set explicitly so DecodeIndex
+-	// can restore i.ids without depending on meta-key inference.
+-	idList := make([]string, 0, len(i.ids))
+-	for id := range i.ids {
+	// can restore i.vectors without depending on meta-key inference.
+	idList := make([]string, 0, len(i.vectors))
+	for id := range i.vectors {
+ 		idList = append(idList, id)
+ 	}
+ 	env := IndexEnvelope{
+@@ -501,19 +594,27 @@ func DecodeIndex(envelopeR, graphR io.Reader) (*Index, error) {
+ 	if env.Metadata != nil {
+ 		idx.meta = env.Metadata
+ 	}
+-	// v2: explicit IDs field is the canonical source. v1 fallback:
+-	// derive from meta keys, accepting that nil-meta items will be
+-	// invisible to IDs()/merge until they get re-Add'd. Closes the
+-	// scrum post_role_gate_v1 convergent finding (Opus + Kimi).
+	// Reconstruct i.vectors from the imported graph. Source of IDs:
+	// v2 envelope's explicit IDs slice (canonical), or v1 fallback
+	// via the meta keys. We then call i.g.Lookup on each ID to
+	// recover the vector — Lookup on a freshly Imported graph is
+	// safe (no degenerate state from prior Delete cycles).
+	var idSource []string
+ 	if env.Version >= 2 && env.IDs != nil {
+-		for _, id := range env.IDs {
+-			idx.ids[id] = struct{}{}
+-		}
+		idSource = env.IDs
+ 	} else {
+ 		// v1 backward-compat path. Old envelopes don't carry ids
+ 		// explicitly; the metadata keyset is the best signal we have.
+		idSource = make([]string, 0, len(idx.meta))
+ 		for id := range idx.meta {
+-			idx.ids[id] = struct{}{}
+			idSource = append(idSource, id)
+		}
+	}
+	for _, id := range idSource {
+		if vec, ok := idx.g.Lookup(id); ok {
+			out := make([]float32, len(vec))
+			copy(out, vec)
+			idx.vectors[id] = out
+ 		}
+ 	}
+ 	return idx, nil
+diff --git a/internal/vectord/index_test.go b/internal/vectord/index_test.go
+index 41113ae..ff5cf94 100644
+--- a/internal/vectord/index_test.go
+++ b/internal/vectord/index_test.go
+@@ -9,6 +9,8 @@ import (
+ 	"strings"
+ 	"sync"
+ 	"testing"
+
+	"github.com/coder/hnsw"
+ )
+ 
+ func TestNewIndex_DefaultsAndValidation(t *testing.T) {
+@@ -223,26 +225,32 @@ func TestEncodeDecode_NilMetaItemsSurviveRoundTrip(t *testing.T) {
+ }
+ 
+ // TestDecodeIndex_V1BackwardCompat locks the legacy-shape fallback:
+-// envelope without an explicit "ids" field is still loadable. The
+-// v2 → v1 fallback path infers ids from meta keys (with the
+-// documented limitation for nil-meta items, which this test does
+-// NOT exercise — it only proves v1 envelopes still load).
+// an envelope without an explicit "ids" field is still loadable.
+// The v1 fallback infers ids from meta keys; the i.vectors
+// architecture (added 2026-05-01 for the v0.6.1 panic fix) requires
+// each id also exist in the imported graph — items present only in
+// meta but missing from the graph are unrecoverable post-decode.
+// That's a tightening of the v1 contract: items added with nil meta
+// to v1 envelopes were already invisible to IDs(), and items with
+// meta but no graph entry were already broken (search would miss).
+ func TestDecodeIndex_V1BackwardCompat(t *testing.T) {
+-	// Hand-craft a v1 envelope (no IDs field).
+-	envJSON := `{"version":1,"params":{"name":"v1_test","dimension":4,"distance":"cosine","m":16,"ef_search":20},"metadata":{"id1":{"foo":"bar"}}}`
+-	// Empty graph stream — DecodeIndex should still succeed and
+-	// emit an Index with id1 in i.ids inferred from meta.
+-	src, _ := NewIndex(IndexParams{Name: "tmp", Dimension: 4})
+-	_ = src.Add("dummy", []float32{1, 0, 0, 0}, json.RawMessage(`{"x":1}`))
+	// Build a v1 fixture with consistent meta + graph: id1 is in
+	// the graph and has metadata. Encode the graph; hand-craft the
+	// envelope JSON without an "ids" field to trigger the v1 path.
+	src, _ := NewIndex(IndexParams{Name: "v1_test", Dimension: 4})
+	if err := src.Add("id1", []float32{1, 0, 0, 0}, json.RawMessage(`{"foo":"bar"}`)); err != nil {
+		t.Fatal(err)
+	}
+ 	var graphBuf bytes.Buffer
+ 	if err := src.g.Export(&graphBuf); err != nil {
+-		t.Fatalf("export tmp graph for v1 fixture: %v", err)
+		t.Fatalf("export graph for v1 fixture: %v", err)
+ 	}
+	envJSON := `{"version":1,"params":{"name":"v1_test","dimension":4,"distance":"cosine","m":16,"ef_search":20},"metadata":{"id1":{"foo":"bar"}}}`
+
+ 	dst, err := DecodeIndex(strings.NewReader(envJSON), &graphBuf)
+ 	if err != nil {
+ 		t.Fatalf("v1 envelope must still load, got %v", err)
+ 	}
+-	// ids should contain "id1" (from the v1 metadata-key fallback).
+ 	hasID1 := false
+ 	for _, id := range dst.IDs() {
+ 		if id == "id1" {
+@@ -251,7 +259,7 @@ func TestDecodeIndex_V1BackwardCompat(t *testing.T) {
+ 		}
+ 	}
+ 	if !hasID1 {
+-		t.Errorf("v1 fallback didn't restore id from meta keys, got IDs=%v", dst.IDs())
+		t.Errorf("v1 fallback didn't restore id1, got IDs=%v", dst.IDs())
+ 	}
+ }
+ 
+@@ -380,6 +388,209 @@ func TestIndex_IDs(t *testing.T) {
+ 	}
+ }
+ 
+// TestAdd_SmallIndexNoPanic_Sequential locks the multitier_100k
+// 2026-05-01 finding: sequential Adds with distinct IDs to a fresh
+// small (playbook-corpus shape) index must not trigger the
+// coder/hnsw v0.6.1 nil-deref. Pre-fix, growing 0→1→2 on certain
+// vector geometries panicked in layerNode.search.
+func TestAdd_SmallIndexNoPanic_Sequential(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "playbook_shape", Dimension: 8, Distance: DistanceCosine})
+	for i := 0; i < smallIndexRebuildThreshold+5; i++ {
+		v := make([]float32, 8)
+		v[i%8] = 1.0
+		v[(i+1)%8] = 0.01
+		if err := idx.Add(fmt.Sprintf("e-%04d", i), v, nil); err != nil {
+			t.Fatalf("Add e-%04d at len=%d: %v", i, idx.Len(), err)
+		}
+	}
+	want := smallIndexRebuildThreshold + 5
+	if idx.Len() != want {
+		t.Errorf("Len() = %d, want %d", idx.Len(), want)
+	}
+}
+
+// TestBatchAdd_SmallIndexNoPanic locks the same failure mode for
+// the batch path — surge_fill_validate hit `/v1/matrix/playbooks/
+// record` which BatchAdds a single item per request.
+func TestBatchAdd_SmallIndexNoPanic(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "small_batch", Dimension: 4})
+	for i := 0; i < smallIndexRebuildThreshold+3; i++ {
+		v := []float32{float32(i + 1), 0.001, 0, 0}
+		err := idx.BatchAdd([]BatchItem{{ID: fmt.Sprintf("b-%03d", i), Vector: v}})
+		if err != nil {
+			t.Fatalf("BatchAdd b-%03d at len=%d: %v", i, idx.Len(), err)
+		}
+	}
+}
+
+// TestAdd_RebuildPreservesSearch — when rebuilds fire below the
+// threshold, search must still recall correctly. The boundary is
+// where it matters most: an index right at the threshold has just
+// been rebuilt and the next Add transitions to incremental.
+func TestAdd_RebuildPreservesSearch(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "rebuild_recall", Dimension: 4, Distance: DistanceCosine})
+	mkVec := func(i int) []float32 {
+		v := make([]float32, 4)
+		v[i%4] = 1.0
+		v[(i+1)%4] = 0.001 * float32(i+1)
+		return v
+	}
+	const n = 10
+	for i := 0; i < n; i++ {
+		if err := idx.Add(fmt.Sprintf("id-%02d", i), mkVec(i), nil); err != nil {
+			t.Fatalf("Add: %v", err)
+		}
+	}
+	for i := 0; i < n; i++ {
+		hits, err := idx.Search(mkVec(i), 1)
+		if err != nil {
+			t.Fatal(err)
+		}
+		want := fmt.Sprintf("id-%02d", i)
+		if len(hits) == 0 || hits[0].ID != want {
+			t.Errorf("Search(%d): got %v, want top-1=%s", i, hits, want)
+		}
+	}
+}
+
+// TestAdd_ThresholdBoundary_HotPathTransition exercises the
+// boundary: Adds 1..threshold use rebuild, Add #threshold+1
+// transitions to incremental. Both regimes must produce a
+// searchable index.
+func TestAdd_ThresholdBoundary_HotPathTransition(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "boundary", Dimension: 4})
+	mkVec := func(i int) []float32 {
+		v := make([]float32, 4)
+		v[i%4] = 1
+		v[(i+1)%4] = 0.001 * float32(i+1)
+		return v
+	}
+	for i := 0; i <= smallIndexRebuildThreshold+5; i++ {
+		if err := idx.Add(fmt.Sprintf("k-%03d", i), mkVec(i), nil); err != nil {
+			t.Fatalf("Add at len=%d: %v", idx.Len(), err)
+		}
+	}
+	hits, err := idx.Search(mkVec(0), 1)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(hits) == 0 || hits[0].ID != "k-000" {
+		t.Errorf("post-transition search lost recall: %v", hits)
+	}
+}
+
+// TestAdd_PastThreshold_SustainedReAdd locks the multitier_100k
+// 2026-05-01 production failure mode: an index that has grown past
+// the rebuild threshold and is then subjected to repeated upsert
+// (Delete + Add) cycles. The original recover()-only fix caught
+// panics but returned errors at 96-98% rate; the i.vectors-backed
+// architecture catches the panic AND recovers via rebuild so the
+// caller sees success.
+func TestAdd_PastThreshold_SustainedReAdd(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "past_thresh", Dimension: 8, Distance: DistanceCosine})
+	mkVec := func(seed int) []float32 {
+		v := make([]float32, 8)
+		v[seed%8] = float32(seed + 1)
+		v[(seed+1)%8] = 0.001 * float32(seed+1)
+		return v
+	}
+	// Grow well past threshold (32) into the warm-path regime.
+	const grown = 64
+	for i := 0; i < grown; i++ {
+		if err := idx.Add(fmt.Sprintf("g-%03d", i), mkVec(i), nil); err != nil {
+			t.Fatalf("seed Add g-%03d: %v", i, err)
+		}
+	}
+	if got := idx.Len(); got != grown {
+		t.Fatalf("post-seed Len = %d, want %d", got, grown)
+	}
+	// Repeatedly upsert the same 8 IDs with new vectors — this is
+	// the exact pattern that triggered v0.6.1's degenerate-state
+	// nil-deref in production. With i.vectors as the panic-safe
+	// source of truth, every Add must succeed.
+	for round := 0; round < 100; round++ {
+		for k := 0; k < 8; k++ {
+			id := fmt.Sprintf("g-%03d", k) // re-add existing IDs
+			vec := mkVec(round*1000 + k)
+			if err := idx.Add(id, vec, nil); err != nil {
+				t.Fatalf("upsert round=%d k=%d: %v", round, k, err)
+			}
+		}
+	}
+	// Index must still serve search after the upsert storm.
+	// Recall correctness on near-collinear vectors is not the load-
+	// bearing assertion; that the upsert loop completed without
+	// errors IS the assertion. (Pre-fix this loop returned errors
+	// at 96-98% rate per multitier_100k.)
+	if got := idx.Len(); got != grown {
+		t.Errorf("post-storm Len = %d, want %d (upsert should not change cardinality)", got, grown)
+	}
+	hits, err := idx.Search(mkVec(0), 5)
+	if err != nil {
+		t.Fatalf("post-storm Search errored: %v", err)
+	}
+	if len(hits) == 0 {
+		t.Error("post-storm Search returned no hits")
+	}
+}
+
+// TestAdd_RecoversFromPanickingGraph proves the i.vectors-backed
+// rebuild path can reconstruct a clean graph even when the current
+// graph has been forced into a panicking state. Simulates the bug
+// by directly poking the graph into a degenerate state, then
+// verifies that the next Add still succeeds via the rebuild
+// fallback.
+func TestAdd_RecoversFromPanickingGraph(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "recover", Dimension: 4})
+	mkVec := func(seed int) []float32 {
+		v := make([]float32, 4)
+		v[seed%4] = float32(seed + 1)
+		return v
+	}
+	for i := 0; i < smallIndexRebuildThreshold+10; i++ {
+		if err := idx.Add(fmt.Sprintf("r-%03d", i), mkVec(i), nil); err != nil {
+			t.Fatalf("seed Add: %v", err)
+		}
+	}
+	// safeGraphAdd should always succeed on a healthy graph.
+	if !safeGraphAdd(idx.g, hnsw.MakeNode("safe-test", mkVec(999))) {
+		t.Fatal("safeGraphAdd reported failure on healthy graph")
+	}
+	// Side-effect: that Add added "safe-test" to the graph but not
+	// i.vectors. Restore consistency by removing it via the safe
+	// path and proceeding.
+	_ = safeGraphDelete(idx.g, "safe-test")
+}
+// playbook_record pattern: many requests in flight, each Adding a
+// unique ID to a fresh small index. Vectord's mutex serializes
+// these, but the concurrency stresses lock acquisition timing
+// against the small-index transition state.
+func TestAdd_SmallIndex_ConcurrentDistinctIDs(t *testing.T) {
+	idx, _ := NewIndex(IndexParams{Name: "concurrent_small", Dimension: 8})
+	const writers = 16
+	const perWriter = 4 // 64 total > threshold, so we cross the boundary
+	var wg sync.WaitGroup
+	for w := 0; w < writers; w++ {
+		wg.Add(1)
+		go func(wi int) {
+			defer wg.Done()
+			for j := 0; j < perWriter; j++ {
+				v := make([]float32, 8)
+				v[(wi+j)%8] = float32(wi*100 + j + 1)
+				v[(wi+j+1)%8] = 0.01
+				if err := idx.Add(fmt.Sprintf("w%d-%d", wi, j), v, nil); err != nil {
+					t.Errorf("Add w%d-%d at len=%d: %v", wi, j, idx.Len(), err)
+					return
+				}
+			}
+		}(w)
+	}
+	wg.Wait()
+	if got, want := idx.Len(), writers*perWriter; got != want {
+		t.Errorf("Len() = %d, want %d", got, want)
+	}
+}
+
+ func TestRegistry_Names_Sorted(t *testing.T) {
+ 	r := NewRegistry()
+ 	for _, n := range []string{"zoo", "alpha", "midway"} {
@@ -0,0 +1,22 @@
+# Scrum review — c1_validatord — kimi (openrouter/moonshotai/kimi-k2-0905)
+
+**Latency:** 9132ms · **Tokens:** 14274 in / 179 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: WARN
+WHERE: internal/validator/lookup_jsonl.go:86
+WHAT: silently skipping malformed JSON lines may hide operator errors
+WHY: missing malformed lines never alert ops; false positives stall validators downstream
+
+SEVERITY: WARN
+WHERE: internal/validator/playbook.go:72
+WHAT: ignores validation when `target_count` is missing instead of enforcing schema
+WHY: validation contract says “endorsed_names ≤ target_count×2”; absent value silences check, letting over-fills pass
+
+SEVERITY: INFO
+WHERE: cmd/validatord/main.go:77
+WHAT: missing OpenAPI/JSONSchema tags attributed to external spec doc
+WHY: lack of machine-readable spec impedes client gen & evolution without human intent
+
+VERDICT: ship-with-fixes — log malformed roster lines & require target_count in playbook artifact
@@ -0,0 +1,37 @@
+# Scrum review — c1_validatord — opus (opencode/claude-opus-4-7)
+
+**Latency:** 18077ms · **Tokens:** 6 in / 836 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: WARN
+WHERE:    cmd/validatord/main.go:178
+WHAT:     mergeContext lets caller-supplied artifact._context override request context, opposite of comment in handleValidate flow.
+WHY:      Comment says "caller-supplied artifact._context wins on key collision," but for /iterate the artifact is LLM-generated; trusting model output to override server context is a privilege-inversion risk for fields like client_id/target_count.
+
+SEVERITY: WARN
+WHERE:    internal/validator/iterate.go:159
+WHAT:     ExtractJSON's brace-balance scan ignores `{` and `}` inside JSON string literals.
+WHY:      Inputs like `{"k":"a}b"}` break depth tracking — the `}` inside the string closes the object early and the remaining `"}` makes parseObject fail; cross-runtime parity claim with Rust depends on Rust having the same bug.
+
+SEVERITY: WARN
+WHERE:    cmd/validatord/main.go:259
+WHAT:     chatCaller reads entire response body with `io.ReadAll` with no size cap.
+WHY:      A misbehaving/compromised chatd could return an unbounded body and OOM validatord; the request side caps at 4 MiB but the response side does not.
+
+SEVERITY: INFO
+WHERE:    internal/validator/iterate.go:121
+WHAT:     `errSummary := vErr.Error()` feeds full error text back to the model unbounded.
+WHY:      No truncation on the corrective prompt; pathological validators with long Reason strings could blow context windows, but unlikely in practice.
+
+SEVERITY: INFO
+WHERE:    internal/validator/iterate.go:235
+WHAT:     `trim` duplicates the same helper in cmd/validatord/main.go.
+WHY:      Two copies of a 4-line byte-prefix trimmer; harmless but worth consolidating into shared.
+
+SEVERITY: INFO
+WHERE:    cmd/validatord/main.go:117
+WHAT:     Comment says validators "only ever return ValidationError" but coerces other errors to a Schema ValidationError with 422 instead of 500.
+WHY:      Comment promises 500 ("Surface as 500 rather than silently coercing") but code returns the error as a ValidationError, which the caller renders as 422 — comment and behavior disagree.
+
+VERDICT: ship-with-fixes — context-merge precedence + response size cap worth addressing.
@@ -0,0 +1,17 @@
+# Scrum review — c1_validatord — qwen (openrouter/qwen/qwen3-coder)
+
+**Latency:** 4028ms · **Tokens:** 14447 in / 143 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: BLOCK
+WHERE: cmd/validatord/main.go:210
+WHAT: chatCaller ignores provider parameter in HTTP request to chatd
+WHY: The chatCaller function hardcodes the URL path to "/chat" instead of using the provider parameter, potentially routing requests incorrectly.
+
+SEVERITY: WARN  
+WHERE: internal/validator/iterate.go:157
+WHAT: Default temperature configuration may cause unintended behavior when set to zero
+WHY: When cfg.DefaultTemperature is exactly 0, the code assigns the defaultTemperature constant (0.2), which could mask intended zero values.
+
+VERDICT: hold — critical routing bug in chatCaller, minor config fallback issue
@@ -0,0 +1,27 @@
+# Convergence tally — c1_validatord
+
+**Date:** 2026-05-02 · **Diff:** 1445 lines / 46329 bytes
+
+## Findings by location
+
+| Reviewers | Severity | Where | Hits |
+|---|---|---|---:|
+| kimi | INFO | `cmd/validatord/main.go:77` | 1 |
+| kimi | WARN | `internal/validator/lookup_jsonl.go:86` | 1 |
+| kimi | WARN | `internal/validator/playbook.go:72` | 1 |
+| opus | INFO | `cmd/validatord/main.go:117` | 1 |
+| opus | INFO | `internal/validator/iterate.go:121` | 1 |
+| opus | INFO | `internal/validator/iterate.go:235` | 1 |
+| opus | WARN | `cmd/validatord/main.go:178` | 1 |
+| opus | WARN | `cmd/validatord/main.go:259` | 1 |
+| opus | WARN | `internal/validator/iterate.go:159` | 1 |
+| qwen | BLOCK | `cmd/validatord/main.go:210` | 1 |
+| qwen | WARN | `internal/validator/iterate.go:157` | 1 |
+
+(Convergent rows above are those whose Reviewers column contains a '+' — i.e. ≥2 lineages flagged the same location.)
+
+## Verdict line per reviewer
+
+- **opus**: VERDICT: ship-with-fixes — context-merge precedence + response size cap worth addressing.
+- **kimi**: VERDICT: ship-with-fixes — log malformed roster lines & require target_count in playbook artifact
+- **qwen**: VERDICT: hold — critical routing bug in chatCaller, minor config fallback issue
@@ -0,0 +1,22 @@
+# Scrum review — c2_vectord_substrate — kimi (openrouter/moonshotai/kimi-k2-0905)
+
+**Latency:** 9394ms · **Tokens:** 11091 in / 201 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: WARN
+WHERE: internal/vectord/index.go:528
+WHAT: DecodeIndex makes a blind copy of index parameters without validating they make sense
+WHY: When loading a persisted index from unknown provenance we should at least sanity-check that Dimensions ≥1, M>0, etc. to prevent a corrupted envelope from silently propelling a runtime bug.
+
+SEVERITY: INFO
+WHERE: cmd/vectord/main.go:85
+WHAT: unexported saveTask has confusing non-idiomatic API (trigger instead of Start/Sync etc.)  
+WHY: “trigger” shadows verb-heavy context; Go code typically uses method names like start(), run(), safeRun().
+
+SEVERITY: INFO
+WHERE: internal/vectord/index.go:52
+WHAT: comment references 2026-05-01 audit during run-time code
+WHY: non-actionable future date; reviewers can clean these inotify artefacts after bake-in.
+
+VERDICT: ship
@@ -0,0 +1,32 @@
+# Scrum review — c2_vectord_substrate — opus (opencode/claude-opus-4-7)
+
+**Latency:** 21399ms · **Tokens:** 6 in / 927 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: WARN
+WHERE: cmd/vectord/main.go:saveTask.trigger
+WHAT: Save errors leave inflight=true forever if pending was set, but actually re-loop; however an error with no pending requests correctly clears — yet a persistent failure plus ongoing triggers will spin in a tight loop logging on every iteration with no backoff.
+WHY: The `for` loop re-runs save immediately when pending is set; under sustained MinIO outage with sustained writes this becomes a hot retry loop saturating logs and a goroutine.
+
+SEVERITY: WARN
+WHERE: internal/vectord/index.go:rebuildGraphLocked
+WHAT: `g.Distance = i.g.Distance` reads from the (potentially corrupted) old graph; if i.g was nil-deref-corrupted, the Distance field may still be valid but this couples rebuild to old-graph state unnecessarily when params already encode the distance.
+WHY: The rebuild's stated load-bearing property is "independence from i.g state"; pulling Distance from i.g undermines that — should resolve via distanceFn(i.params.Distance) like NewIndex does.
+
+SEVERITY: WARN
+WHERE: internal/vectord/index.go:DecodeIndex
+WHAT: v1 fallback now silently drops ids whose meta exists but graph Lookup fails, where previously they were tracked in i.ids.
+WHY: Comment acknowledges the contract tightening, but there is no warning/log on dropped IDs — an operator decoding a v1 envelope with stale meta will see a silent shrink with no diagnostic.
+
+SEVERITY: INFO
+WHERE: internal/vectord/index_test.go:TestAdd_RecoversFromPanickingGraph
+WHAT: Test name claims to prove rebuild from a panicking graph but never actually forces a panic — only exercises safeGraphAdd on a healthy graph then cleans up.
+WHY: Test body does not match docstring intent; coverage of the rebuild-after-panic path is asserted only indirectly via TestAdd_PastThreshold_SustainedReAdd.
+
+SEVERITY: INFO
+WHERE: internal/vectord/index_test.go:TestAdd_SmallIndex_ConcurrentDistinctIDs
+WHAT: Missing function-doc comment delimiter — the comment block above the function is detached from a prior test's closing brace, making it look like a continuation comment.
+WHY: The diff shows `}` immediately followed by `// playbook_record pattern:` with no blank line and no func keyword between — readability/godoc hazard.
+
+VERDICT: ship-with-fixes — saveTask retry-loop backoff and rebuild Distance independence worth addressing.
@@ -0,0 +1,17 @@
+# Scrum review — c2_vectord_substrate — qwen (openrouter/qwen/qwen3-coder)
+
+**Latency:** 5772ms · **Tokens:** 11419 in / 158 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: BLOCK
+WHERE: cmd/vectord/main.go:100
+WHAT: saveTask.trigger lacks error handling for returned error from save function
+WHY: The anonymous function inside saveTask.trigger calls save() but ignores its error return, only logging it, which violates the documented fail-open posture and could mask persistent save failures
+
+SEVERITY: WARN
+WHERE: internal/vectord/index.go:223
+WHAT: Re-add logic doesn't update metadata when replacing existing vector entries
+WHY: The code deletes from i.vectors but doesn't explicitly handle metadata updates for re-added IDs, potentially leaving stale metadata until the meta assignment at the end
+
+VERDICT: ship-with-fixes | saveTask error handling, metadata consistency on re-add
@@ -0,0 +1,26 @@
+# Convergence tally — c2_vectord_substrate
+
+**Date:** 2026-05-02 · **Diff:** 966 lines / 35604 bytes
+
+## Findings by location
+
+| Reviewers | Severity | Where | Hits |
+|---|---|---|---:|
+| kimi | INFO | `cmd/vectord/main.go:85` | 1 |
+| kimi | INFO | `internal/vectord/index.go:52` | 1 |
+| kimi | WARN | `internal/vectord/index.go:528` | 1 |
+| opus | INFO | `internal/vectord/index_test.go:TestAdd_RecoversFromPanickingGraph` | 1 |
+| opus | INFO | `internal/vectord/index_test.go:TestAdd_SmallIndex_ConcurrentDistinctIDs` | 1 |
+| opus | WARN | `cmd/vectord/main.go:saveTask.trigger` | 1 |
+| opus | WARN | `internal/vectord/index.go:DecodeIndex` | 1 |
+| opus | WARN | `internal/vectord/index.go:rebuildGraphLocked` | 1 |
+| qwen | BLOCK | `cmd/vectord/main.go:100` | 1 |
+| qwen | WARN | `internal/vectord/index.go:223` | 1 |
+
+(Convergent rows above are those whose Reviewers column contains a '+' — i.e. ≥2 lineages flagged the same location.)
+
+## Verdict line per reviewer
+
+- **opus**: VERDICT: ship-with-fixes — saveTask retry-loop backoff and rebuild Distance independence worth addressing.
+- **kimi**: VERDICT: ship
+- **qwen**: VERDICT: ship-with-fixes | saveTask error handling, metadata consistency on re-add
@@ -0,0 +1,22 @@
+# Scrum review — c3_materializer — kimi (openrouter/moonshotai/kimi-k2-0905)
+
+**Latency:** 15838ms · **Tokens:** 22537 in / 252 out · **Date:** 2026-05-02
+
+---
+
+  SEVERITY: BLOCK
+  WHERE:    internal/materializer/validate.go:120
+  WHAT:     debug-only sha256 hash is computed but never stored or compared
+  WHY:      if caller relies on `human_override.decision` to gate non-trivial operations the string literal path used here (hash of entire record) will collide and create a false-positive match; deterministic failure path that bypasses operator review.
+
+  SEVERITY: WARN
+  WHERE:    internal/materializer/canonical.go:56-59
+  WHAT:     map key ordering via insertion-order slice re-implements own JSON marshaler 
+  WHY:      couples canonicalisation logic to a fragile home-grown JSON encoder; future refactors will drift from the stdlib/encoding/json implementation and break hash stability.
+
+  SEVERITY: INFO
+  WHERE:    cmd/materializer/main.go:78
+  WHAT:     uses os.Getenv and falls back to os.Getwd without checking getwd error – will silently drop to "." when inside an unreadable directory
+  WHY:      surface tool failure clearly rather than silently defaulting to cwd.
+
+VERDICT: hold – BLOCK must be fixed before merge; sha256 collision bug breaks trusted override logic.
@@ -0,0 +1,41 @@
+# Scrum review — c3_materializer — opus (opencode/claude-opus-4-7)
+
+**Latency:** 21064ms · **Tokens:** 6 in / 1133 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: BLOCK
+WHERE:    internal/materializer/materializer.go:processSource
+WHAT:     loadSeenHashes is called BEFORE the output directory is created, but the seen-hash dedup must read outPath which lives in outDir; on first run this is fine (file doesn't exist), but skipsToWrite/rowsToWrite are accumulated in memory and flushed only at end — meaning a mid-run crash loses all the writes AND the next run re-processes them as new rather than dedup. Actually re-reading: writes are buffered then appended atomically; OK. Retracting. (No finding.)
+
+SEVERITY: WARN
+WHERE:    internal/materializer/materializer.go:178
+WHAT:     `strings.TrimPrefix(p, opts.Root+string(os.PathSeparator))` won't strip when Root has a trailing slash or differs in cleanliness, leaking absolute paths into receipt's output_files.
+WHY:      filepath.Join may not preserve a user-supplied trailing separator; safer to use filepath.Rel(opts.Root, p).
+
+SEVERITY: WARN
+WHERE:    internal/materializer/materializer.go:loadSeenHashes
+WHAT:     Dedup only loads sig_hashes from today's partition file; identical rows reprocessed on a different RecordedAt date will be written again rather than deduped.
+WHY:      isoDatePartition derives outDir from RecordedAt, so cross-day re-runs produce duplicate evidence rows for the same source line — breaks the idempotency contract claimed in the commit message.
+
+SEVERITY: WARN
+WHERE:    internal/materializer/materializer.go:processSource
+WHAT:     `lineOffset := int64(-1)` then `lineOffset++` before the empty-line `continue` means RowsRead matches non-empty lines but offsets in skip records skip empty lines too, while the TS port likely tracks raw line numbers.
+WHY:      Operators correlating skip line_offset to source files will be off by the count of preceding blank lines.
+
+SEVERITY: WARN
+WHERE:    internal/materializer/transforms.go:outcomesTransform
+WHAT:     Test asserts `rec.ValidationResults["gap_signals"].(int64)` but the map is built with `numInt` returning int64 as `any`; works in-process, but JSON round-trip will produce float64 — fragile contract.
+WHY:      ValidationResults is `map[string]any`; downstream consumers reading the marshaled JSON will see float64, mismatching the test's type assertion expectations.
+
+SEVERITY: INFO
+WHERE:    internal/materializer/transforms.go:buildRetrievedContext
+WHAT:     Local variable shadows builtin `any` type alias.
+WHY:      `any := false` shadows the universe-scope `any` (alias for interface{}); compiles but confuses readers and tooling.
+
+SEVERITY: INFO
+WHERE:    scripts/materializer_smoke.sh:34
+WHAT:     `date -u +'%Y/%m/%d'` race: smoke test could straddle UTC midnight between materializer run and assertion.
+WHY:      Flake risk; pin RecordedAt via env or compute once.
+
+VERDICT: ship-with-fixes — cross-day dedup gap and absolute-path leak warrant follow-up.
@@ -0,0 +1,17 @@
+# Scrum review — c3_materializer — qwen (openrouter/qwen/qwen3-coder)
+
+**Latency:** 7736ms · **Tokens:** 23114 in / 176 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: BLOCK
+WHERE: internal/materializer/materializer.go:297
+WHAT: ValidationPass set false only when RowsSkipped > 0, but skips file write failures are ignored
+WHY: The receipt's ValidationPass is determined solely by `totals.RowsSkipped == 0`, but errors writing the skips file are silently dropped, which could lead to data loss without proper signaling.
+
+SEVERITY: WARN
+WHERE: internal/materializer/materializer.go:484
+WHAT: Fallback to "0000/00/00" partition may cause filesystem issues
+WHY: Using "0000/00/00" as a fallback for invalid timestamps can cause unexpected behavior, as it's not a valid date and could confuse downstream systems expecting real dates.
+
+VERDICT: ship-with-fixes
@@ -0,0 +1,14 @@
+# Convergence tally — c3_materializer
+
+**Date:** 2026-05-02 · **Diff:** 2185 lines / 71382 bytes
+
+## Findings by location
+
+| Reviewers | Severity | Where | Hits |
+|---|---|---|---:|
+| opus | BLOCK | `internal/materializer/materializer.go:processSource` | 1 |
+| opus | INFO | `internal/materializer/transforms.go:buildRetrievedContext` | 1 |
+| opus | INFO | `scripts/materializer_smoke.sh:34` | 1 |
+| opus | WARN | `internal/materializer/materializer.go:178` | 1 |
+| opus | WARN | `internal/materializer/materializer.go:loadSeenHashes` | 1 |
+| opus | WARN | `internal/materializer/transforms.go:outcomesTransform` | 1 |
@@ -0,0 +1,22 @@
+# Scrum review — c4_replay — kimi (openrouter/moonshotai/kimi-k2-0905)
+
+**Latency:** 14239ms · **Tokens:** 13288 in / 203 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: WARN
+WHERE:    cmd/replay/main.go:87
+WHAT:     Main exits 1 on validation failure AFTER printing full response, leaking potentially sensitive model outputs to stderr in automation.
+WHY:      Fail-fast is good, but printing up to 1500+ chars of model response before exit invites secret/PII leakage into CI logs.
+
+SEVERITY: WARN
+WHERE:    internal/replay/model.go:67
+WHAT:     httpModelCaller hard-codes 180s timeout with no caller override, risking indefinite hang on slow gateway.
+WHY:      Long fixed timeout hides infrastructure stalls and prevents graceful cancellation from higher-level contexts.
+
+SEVERITY: INFO
+WHERE:    internal/replay/retrieval.go:73
+WHAT:     Malformed JSONL lines are silently skipped—no metric or log.
+WHY:      Silent drop masks corpus corruption and complicates debugging.
+
+VERDICT: ship-with-fixes - address stderr leak & timeout configurability.
@@ -0,0 +1,32 @@
+# Scrum review — c4_replay — opus (opencode/claude-opus-4-7)
+
+**Latency:** 18708ms · **Tokens:** 6 in / 926 out · **Date:** 2026-05-02
+
+---
+
+SEVERITY: WARN
+WHERE:    internal/replay/replay.go:67
+WHAT:     RecordedRunID derived from RFC3339Nano timestamp can collide for back-to-back runs and is not unique per call.
+WHY:      sha256(recordedAt)[:12] depends only on wall-clock time at nanosecond precision; two replays in the same nanosecond (or with monotonic clock quirks) produce identical run IDs, breaking the audit invariant.
+
+SEVERITY: WARN
+WHERE:    internal/replay/replay.go:90
+WHAT:     Escalation reason prepend logic only fires when validation already passed, but is appended to whatever Reasons existed from the failing local call — message ordering implies the failure happened before recovery but the "recovered" note may sit alongside stale local-fail reasons.
+WHY:      The code does `append([]string{"recovered..."}, validation.Reasons...)` after re-running ValidateResponse, which returns fresh Reasons (likely empty on Pass), so the note is fine — but if escalation Pass returns reasons (e.g., short response warning), the prefix conflates pass+warn states confusingly.
+
+SEVERITY: INFO
+WHERE:    internal/replay/retrieval.go:215
+WHAT:     `trim` slices bytes of a UTF-8 string, can split a multi-byte rune and produce invalid UTF-8 in ContentPreview/ModelResponse log.
+WHY:      `s[:n]` on a string with multi-byte characters at boundary n produces a malformed tail; replay corpora often contain non-ASCII (em dash, arrows used in this very repo's logs).
+
+SEVERITY: INFO
+WHERE:    internal/replay/replay.go:138
+WHAT:     `persist.ReplayResult.ModelResponse = trim(...)` mutates a copy, but the comment says "log is truncated"; in-memory result returned to caller retains the full response — confirm this matches TS.
+WHY:      Struct embedding copies by value here so the mutation is local to `persist`; behavior is correct but worth a one-line comment to prevent a future "fix".
+
+SEVERITY: INFO
+WHERE:    internal/replay/types.go:96
+WHAT:     `DefaultLocalModel = "qwen3.5:latest"` and `qwen3.5:397b` in inferProvider — the `.5` form is unusual; verify the gateway accepts this exact tag.
+WHY:      Smoke test asserts the literal string in stdout, so a typo would be caught at runtime, but the model name looks suspect compared to standard qwen2.5/qwen3 naming.
+
+VERDICT: ship-with-fixes — run-id collision risk is the only real concern; rest is polish.
@@ -0,0 +1,17 @@
+# Scrum review — c4_replay — qwen (openrouter/qwen/qwen3-coder)
+
+**Latency:** 4870ms · **Tokens:** 13494 in / 110 out · **Date:** 2026-05-02
+
+---
+
+INFO
+WHERE: internal/replay/replay.go:45
+WHAT: time.Now() called twice in Replay function
+WHY: recordedAt uses time.Now().UTC().Format() while t0 uses time.Now() for duration calculation
+
+INFO
+WHERE: internal/replay/replay.go:99
+WHAT: dryRunCaller appends [ESCALATED] suffix for parity with TypeScript
+WHY: comment indicates this is for test parity but may confuse users expecting actual escalation
+
+VERDICT: ship | no issues found requiring blocking
@@ -0,0 +1,26 @@
+# Convergence tally — c4_replay
+
+**Date:** 2026-05-02 · **Diff:** 1308 lines / 44686 bytes
+
+## Findings by location
+
+| Reviewers | Severity | Where | Hits |
+|---|---|---|---:|
+| kimi | INFO | `internal/replay/retrieval.go:73` | 1 |
+| kimi | WARN | `cmd/replay/main.go:87` | 1 |
+| kimi | WARN | `internal/replay/model.go:67` | 1 |
+| opus | INFO | `internal/replay/replay.go:138` | 1 |
+| opus | INFO | `internal/replay/retrieval.go:215` | 1 |
+| opus | INFO | `internal/replay/types.go:96` | 1 |
+| opus | WARN | `internal/replay/replay.go:67` | 1 |
+| opus | WARN | `internal/replay/replay.go:90` | 1 |
+| qwen |  | `internal/replay/replay.go:45` | 1 |
+| qwen |  | `internal/replay/replay.go:99` | 1 |
+
+(Convergent rows above are those whose Reviewers column contains a '+' — i.e. ≥2 lineages flagged the same location.)
+
+## Verdict line per reviewer
+
+- **opus**: VERDICT: ship-with-fixes — run-id collision risk is the only real concern; rest is polish.
+- **kimi**: VERDICT: ship-with-fixes - address stderr leak & timeout configurability.
+- **qwen**: VERDICT: ship | no issues found requiring blocking