feat(panel): N3 — StrategyDrift panel scaffold (config + README)

New clade panel 'strategy-drift' for visualising the learning loop's
strategy recommendations and drift events. Scaffolded with config
(K9 kennel format matching other panels like verisim-feeds) + README
documenting the four data sources and four views:

  - Timeline: per-class top-prover trajectory over time
  - Certs Grid: 11-class × 42-prover cert status grid (PROVEN/
    pending/sanctified borders)
  - Drift Events: append-only log of shift events with candidate
    re-queue counts
  - Coverage Bars: per-class n_repos × n_provers × n_attempts

Data sources:
  - verisim-api /strategy, /certificates, /coverage endpoints
  - hypatia Hypatia.Rules.StrategyDrift ETS snapshot

Backend endpoints all live and tested (committed in verisimdb
e7c5e5a, b660778, a4b5453 and hypatia d6b5144). ReScript/Deno
panel module itself is scheduled as follow-up work.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: hyperpolymath

panel-clades/clades/strategy-drift/README.adoc

@@ -0,0 +1,74 @@
+// SPDX-License-Identifier: PMPL-1.0-or-later
+= Strategy Drift Panel (StrategyDrift)
+:toc: preamble
+
+Visualises the learning loop's strategy recommendations over time,
+the PROVEN/SANCTIFY certificate landscape, and strategy-shift events
+that trigger re-proving of previously-failed attempts.
+
+== Data Sources
+
+Four endpoints on `verisim-api` plus one ETS snapshot on `hypatia`:
+
+[cols="1,2,2"]
+|===
+| Source | Endpoint | Refresh
+
+| Strategy
+| `GET /api/v1/proof_attempts/strategy?class=X`
+| On user action (class selector)
+
+| Certificates
+| `GET /api/v1/proof_attempts/certificates`
+| Every 60s
+
+| Coverage
+| `GET /api/v1/proof_attempts/coverage`
+| Every 60s
+
+| Drift snapshot
+| `Hypatia.Rules.StrategyDrift.snapshot/0` (ETS via HTTP bridge)
+| Every 60s
+|===
+
+== Views
+
+=== Timeline
+Per-class top-prover trajectory. Each obligation class gets its own row;
+colour-coded segments show which prover held the top rank during that
+time window. Drift events appear as vertical separators.
+
+=== Certs Grid
+11-obligation-classes × 42-provers cell grid. Each cell shows
+`success_rate × total_attempts` as size and colour intensity. Cell
+border indicates certificate status:
+
+* thin grey — pending
+* green — proven
+* gold double-border — proven AND part of a sanctified class
+
+=== Drift Events
+Append-only log of `{timestamp, class, old_top, new_top, candidates_requeued}`
+records, produced by `Hypatia.Rules.StrategyDrift.check_all_shifts/1`
+on each polling tick.
+
+=== Coverage Bars
+Per-class horizontal bar triple: `n_repos`, `n_provers`, `n_attempts`.
+Surfaces how narrow or broad the evidence base is for each class —
+a class with `n_repos=1` has narrow cross-repo transfer confidence.
+
+== Related
+
+* verisimdb source — `.../proof_attempts*` handlers and MV DDL
+* hypatia — `Hypatia.Rules.ProofStrategySelection` (N1 novelty gate +
+  N2 quarantine) and `Hypatia.Rules.StrategyDrift` (N4 drift tracker)
+* `docs/design/DESIGN-2026-04-05-V4-proof-attempts-vqldt.adoc` (V4 cert design)
+
+== Implementation Status (2026-04-05)
+
+* Config scaffolded (this file)
+* Backend endpoints live on `verisim-api` (strategy, certificates,
+  coverage — all tested, plus hypatia ETS snapshot via StrategyDrift)
+* ReScript/Deno panel module — not yet written (placeholder for
+  follow-up sprint). Skeleton files follow the pattern established by
+  `verisim-feeds` and other real panels.

panel-clades/clades/strategy-drift/config.k9.ncl

@@ -0,0 +1,71 @@
+# K9 Kennel: Strategy Drift Panel Configuration
+# SPDX-License-Identifier: PMPL-1.0-or-later
+#
+# N3: Panel showing top-prover-per-class over time, with drift events
+# flagged. Data source: Verisim's /api/v1/proof_attempts/strategy and
+# /certificates endpoints, plus hypatia's StrategyDrift ETS snapshots.
+{
+  pedigree = {
+    name = "strategy-drift-config",
+    version = "0.1.0",
+    author = "Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>",
+    leash = 'Kennel,
+  },
+  panel = {
+    id = "StrategyDrift",
+    short_name = "SD",
+    description = "Top-prover-per-class over time + PROVEN/SANCTIFY certificates + drift events",
+    has_backend = true,
+    clade_id = "strategy-drift",
+    isolation_tier = "native",
+  },
+  wiring = {
+    model_slice = "strategyDrift",
+    msg_namespace = "strategyDriftMsg",
+    view_route = "PanelStrategyDrift",
+    module_name = "StrategyDrift",
+  },
+  data_sources = {
+    primary = {
+      endpoint = "http://localhost:8080/api/v1/proof_attempts/strategy",
+      method = "GET",
+      query_params = ["class", "limit"],
+    },
+    certificates = {
+      endpoint = "http://localhost:8080/api/v1/proof_attempts/certificates",
+      method = "GET",
+      query_params = ["class", "prover", "evidence_limit"],
+    },
+    coverage = {
+      endpoint = "http://localhost:8080/api/v1/proof_attempts/coverage",
+      method = "GET",
+    },
+    drift_snapshot = {
+      source = "hypatia.Hypatia.Rules.StrategyDrift.snapshot/0",
+      refresh_interval_seconds = 60,
+    },
+  },
+  views = {
+    timeline = {
+      description = "Per-class top prover trajectory. X: time. Y: class. Colour: prover.",
+      obligation_classes = [
+        "safety", "linearity", "termination", "equiv", "correctness",
+        "confluence", "totality", "invariant", "refinement", "model-check", "other"
+      ],
+    },
+    certs_grid = {
+      description = "11-class × N-prover grid. Cell = success_rate × total_attempts. Border = cert status (proven/pending/sanctified).",
+    },
+    drift_events = {
+      description = "List of (timestamp, class, old_top, new_top, candidates_requeued) shifts.",
+    },
+    coverage_bars = {
+      description = "Per-class: n_repos × n_provers × n_attempts horizontal bars.",
+    },
+  },
+  obligations = {
+    requires_tests = true,
+    requires_backend_health = true,
+    requires_clade_manifest = true,
+  },
+}