instar 1.3.216 → 1.3.218

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "instar",
-  "version": "1.3.216",
+  "version": "1.3.218",
   "description": "Coherence infrastructure for self-evolving AI agents — on the Claude Code or Codex subscription you already have.",
   "type": "module",
   "main": "dist/index.js",

package/src/data/builtin-manifest.json

@@ -1,8 +1,8 @@
 {
   "$schema": "./builtin-manifest.schema.json",
   "schemaVersion": 1,
-  "generatedAt": "2026-06-03T09:17:11.627Z",
-  "instarVersion": "1.3.216",
+  "generatedAt": "2026-06-03T10:21:44.524Z",
+  "instarVersion": "1.3.218",
   "entryCount": 195,
   "entries": {
     "hook:session-start": {
@@ -418,7 +418,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:agents": {
@@ -426,7 +426,7 @@
       "type": "route-group",
       "domain": "sessions",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:backups": {
@@ -434,7 +434,7 @@
       "type": "route-group",
       "domain": "operations",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:git": {
@@ -442,7 +442,7 @@
       "type": "route-group",
       "domain": "coordination",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:memory": {
@@ -450,7 +450,7 @@
       "type": "route-group",
       "domain": "memory",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:semantic": {
@@ -458,7 +458,7 @@
       "type": "route-group",
       "domain": "memory",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:status": {
@@ -466,7 +466,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:capabilities": {
@@ -474,7 +474,7 @@
       "type": "route-group",
       "domain": "mapping",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:project-map": {
@@ -482,7 +482,7 @@
       "type": "route-group",
       "domain": "mapping",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:coherence": {
@@ -490,7 +490,7 @@
       "type": "route-group",
       "domain": "coherence",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:topic-bindings": {
@@ -498,7 +498,7 @@
       "type": "route-group",
       "domain": "sessions",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:context": {
@@ -506,7 +506,7 @@
       "type": "route-group",
       "domain": "context",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:scope-coherence": {
@@ -514,7 +514,7 @@
       "type": "route-group",
       "domain": "coherence",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:canonical-state": {
@@ -522,7 +522,7 @@
       "type": "route-group",
       "domain": "state",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:ci": {
@@ -530,7 +530,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:sessions": {
@@ -538,7 +538,7 @@
       "type": "route-group",
       "domain": "sessions",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:jobs": {
@@ -546,7 +546,7 @@
       "type": "route-group",
       "domain": "scheduling",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:skip-ledger": {
@@ -554,7 +554,7 @@
       "type": "route-group",
       "domain": "scheduling",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:telegram": {
@@ -562,7 +562,7 @@
       "type": "route-group",
       "domain": "communication",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:attention": {
@@ -570,7 +570,7 @@
       "type": "route-group",
       "domain": "communication",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:relationships": {
@@ -578,7 +578,7 @@
       "type": "route-group",
       "domain": "relationships",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:feedback": {
@@ -586,7 +586,7 @@
       "type": "route-group",
       "domain": "feedback",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:updates": {
@@ -594,7 +594,7 @@
       "type": "route-group",
       "domain": "updates",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:dispatches": {
@@ -602,7 +602,7 @@
       "type": "route-group",
       "domain": "dispatches",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:quota": {
@@ -610,7 +610,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:publishing": {
@@ -618,7 +618,7 @@
       "type": "route-group",
       "domain": "publishing",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:private-views": {
@@ -626,7 +626,7 @@
       "type": "route-group",
       "domain": "publishing",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:tunnel": {
@@ -634,7 +634,7 @@
       "type": "route-group",
       "domain": "networking",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:events": {
@@ -642,7 +642,7 @@
       "type": "route-group",
       "domain": "networking",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:evolution": {
@@ -650,7 +650,7 @@
       "type": "route-group",
       "domain": "evolution",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:watchdog": {
@@ -658,7 +658,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:topic-memory": {
@@ -666,7 +666,7 @@
       "type": "route-group",
       "domain": "memory",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:state-sync": {
@@ -674,7 +674,7 @@
       "type": "route-group",
       "domain": "coordination",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:intent": {
@@ -682,7 +682,7 @@
       "type": "route-group",
       "domain": "intent",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:triage": {
@@ -690,7 +690,7 @@
       "type": "route-group",
       "domain": "safety",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:operations": {
@@ -698,7 +698,7 @@
       "type": "route-group",
       "domain": "safety",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:sentinel": {
@@ -706,7 +706,7 @@
       "type": "route-group",
       "domain": "safety",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:trust": {
@@ -714,7 +714,7 @@
       "type": "route-group",
       "domain": "safety",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:monitoring": {
@@ -722,7 +722,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:commitments": {
@@ -730,7 +730,7 @@
       "type": "route-group",
       "domain": "commitments",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:episodes": {
@@ -738,7 +738,7 @@
       "type": "route-group",
       "domain": "memory",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:messages": {
@@ -746,7 +746,7 @@
       "type": "route-group",
       "domain": "coordination",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:system-reviews": {
@@ -754,7 +754,7 @@
       "type": "route-group",
       "domain": "monitoring",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "route-group:machine-mesh": {
@@ -770,7 +770,7 @@
       "type": "route-group",
       "domain": "security",
       "sourcePath": "src/server/routes.ts",
-      "contentHash": "bd2065782b05f1eb6317fd27b95974384974f9534052ec527ccded39153e4a88",
+      "contentHash": "3ebec02e2055c35359e8c316f2516e427007a10f3858c987edae682833276aa6",
       "since": "2025-01-01"
     },
     "cli:init": {
@@ -1474,7 +1474,7 @@
       "type": "subsystem",
       "domain": "server",
       "sourcePath": "src/server/AgentServer.ts",
-      "contentHash": "a251a9d664b4ba0fc86a474813a315da4a3fd2171b2f0d82e95bf1c12ec1af88",
+      "contentHash": "28ca6d76d81717a42a58dd6d72237158a511f72e6764a43eecfeb46e81da0d90",
       "since": "2025-01-01"
     },
     "subsystem:session-manager": {

package/upgrades/1.3.217.md

@@ -0,0 +1,41 @@
+# Upgrade Guide — vNEXT
+
+<!-- assembled-by: assemble-next-md -->
+<!-- bump: minor -->
+
+## What Changed
+
+Adds `ApprenticeshipCycleSlaMonitor`, an observe-only signal for apprenticeship differential cycles
+that have stayed open longer than the configured SLA. The monitor reads `ApprenticeshipCycleStore`,
+filters to open cycles, computes age from `createdAt`, and flags cycles older than the configured
+overdue threshold, which defaults to 120 minutes.
+
+The config gate ships off by default. When enabled, the server wires the signal into the existing
+`TokenLedgerPoller` cadence via an after-tick hook, so this PR does not add another timer. The
+monitor can raise attention, but it dedupes by cycle id in memory and never mutates the cycle store.
+
+Adds `GET /apprenticeship/cycles/overdue?instanceId=` as a read-only authenticated route. It returns
+`{ "overdue": [{ "id", "instanceId", "cycleNumber", "ageMinutes", "createdAt" }] }` and returns 503
+when the monitor is unavailable or disabled.
+
+## What to Tell Your User
+
+- **Overdue apprenticeship cycles**: "I now have a quiet way to spot mentoring cycles that have been
+  left open too long. It starts turned off, and if you ask me to enable it, it can point out stale
+  cycles without changing or closing them."
+
+## Summary of New Capabilities
+
+| Capability | How to Use |
+|-----------|-----------|
+| Overdue apprenticeship cycle signal | Enable the apprenticeship cycle SLA monitor when you want stale open cycles surfaced. |
+| Read overdue cycles | `GET /apprenticeship/cycles/overdue?instanceId=` returns the overdue set when the monitor is enabled. |
+
+## Evidence
+
+Verification:
+
+- Unit: injected clock/store coverage for open vs closed cycles, age thresholding, disabled mode,
+  and per-cycle attention dedup across ticks.
+- Integration: route auth, disabled 503, and instance-scoped overdue results.
+- E2E: `AgentServer` initializes the enabled monitor and serves the overdue route as 200, not 503.

package/upgrades/1.3.218.md

@@ -0,0 +1,42 @@
+# Upgrade Guide — vNEXT
+
+<!-- assembled-by: assemble-next-md -->
+<!-- bump: patch -->
+
+## What Changed
+
+Fixes the LLM circuit breaker so it honors a provider's self-reported reset window when the
+provider phrases it as "reset **after** N seconds" — not just "reset **in** N seconds".
+
+`classifyRateLimit`'s `parseRetryAfterMs` previously matched only `resets in <N>s` / `try again in
+<N>s`. Gemini reports quota exhaustion as **"Your quota will reset after 8s"**, which did not parse
+→ `retryAfterMs` came back `undefined` → the breaker fell back to the blunt `DEFAULT_OPEN_MS`
+(15 minutes). The practical effect, observed live on the `gemini-cli` agent (2026-06-03): an
+8-second provider reset triggered a **900-second global LLM pause** — a ~110× over-correction that
+repeatedly froze the entire agent on a limit that had already cleared.
+
+The fix extends the two duration patterns from `\s+in\s+` to `\s+(?:in|after)\s+`, so "reset after
+Ns/Nm" now parses and the breaker opens for the floored provider window (min 30s) instead of the
+flat 15-minute default. No behavior change for the existing "reset in" phrasing.
+
+## What to Tell Your User
+
+- **Faster recovery from short rate limits**: "When a provider tells me its quota resets in a few
+  seconds, I now wait that short window instead of pausing all LLM work for a blunt 15 minutes. If
+  you saw a Gemini-backed agent freeze for long stretches after a brief rate limit, that's fixed."
+
+## Summary of New Capabilities
+
+| Capability | How to Use |
+|-----------|-----------|
+| Honors "reset after Ns" windows | Automatic — the circuit breaker now parses Gemini-style "reset after N" hints and opens for that window (floored to 30s) instead of the 15-minute default. |
+
+## Evidence
+
+Verification:
+
+- Unit: `classifyRateLimit('…quota will reset after 8s')` → `retryAfterMs === 8000`; `reset after 5
+  minutes` → `300000`; existing "reset in" cases unchanged.
+- Unit (end-to-end regression): the real Gemini message classified + fed to `onRateLimited` opens
+  the breaker for the floored ~30s window and recovers there, NOT at 900s.
+- Full `llm-circuit-breaker-wait` + `LlmCircuitBreaker` suites green (61 tests); `tsc --noEmit` clean.

package/upgrades/apprenticeship-cycle-sla.eli16.md

@@ -0,0 +1,37 @@
+# Apprenticeship overdue-cycle SLA signal
+
+The apprenticeship system now has a quiet way to notice when a mentoring cycle has been left open
+too long.
+
+A cycle is one mentor/overseer loop: task, mentee output, mentor notes, overseer differences, and
+coaching. Those cycles already live in SQLite. This change adds a monitor that only reads them. If a
+cycle is still open and its `createdAt` timestamp is older than the configured limit, the monitor
+marks it overdue.
+
+The default limit is 120 minutes, but the whole signal ships turned off:
+
+```json
+{
+  "monitoring": {
+    "apprenticeshipCycleSla": {
+      "enabled": false,
+      "overdueAfterMinutes": 120
+    }
+  }
+}
+```
+
+When someone turns it on, it can raise an attention item once for each overdue cycle. It remembers
+which cycle ids it already raised during the current server process, so it does not ping every tick
+for the same stale cycle.
+
+There is also a read-only API route:
+
+`GET /apprenticeship/cycles/overdue?instanceId=...`
+
+That route returns the overdue cycle ids, instance ids, cycle numbers, age in minutes, and creation
+timestamps. If the monitor is disabled or unavailable, the route returns 503 instead of pretending it
+is active.
+
+The important boundary: this is only a signal. It does not close cycles, change cycle records, or
+decide that the work was bad. It only says, "this open cycle has been open longer than expected."

package/upgrades/circuit-breaker-reset-after.eli16.md

@@ -0,0 +1,23 @@
+# Circuit Breaker "Reset After" Fix — ELI16
+
+Viewable private artifact:
+https://echo.dawn-tunnel.dev/view/bfae9763-aa22-4486-bc7d-9b7b2e470658?sig=72ade47ad66400b09f8ef262b0a5055fbf5f099684c0b45bb2da1b08b6977743
+
+When an AI provider like Gemini runs low on capacity, it usually tells you when to come back:
+"your quota will reset after 8 seconds." Instar has a safety switch — a **circuit breaker** — that
+briefly pauses AI calls after a rate limit so it doesn't keep hammering an overloaded provider.
+
+The breaker is smart enough to read the provider's "come back in N seconds" hint and only pause that
+long. But it was only listening for the phrase **"reset IN N seconds."** Gemini says **"reset AFTER
+N seconds."** That one-word difference meant Instar couldn't read Gemini's hint at all — so it used
+its blunt fallback: pause **everything for 15 minutes.**
+
+The result: Gemini would say "I'll be ready in 8 seconds," and Instar would freeze the whole agent
+for **15 minutes** — about **110× longer** than needed. On the live Gemini agent this happened over
+and over, making it look stuck.
+
+The fix teaches the breaker to also understand **"reset after N."** Now an 8-second reset means
+roughly a 30-second pause (a sensible floor), not 15 minutes — so the agent recovers quickly instead
+of sitting idle.
+
+_Tier-1 fix · branch `echo/circuit-breaker-reset-after` · 3 unit tests + full breaker suites green._

package/upgrades/side-effects/apprenticeship-cycle-sla.md

@@ -0,0 +1,17 @@
+### apprenticeship-cycle-sla side effects
+
+Risk reviewed: overdue-cycle attention spam.
+
+Mitigation:
+
+- The feature ships disabled by default at `monitoring.apprenticeshipCycleSla.enabled = false`.
+- When enabled, `ApprenticeshipCycleSlaMonitor` records raised cycle ids in memory and raises at
+  most one attention item per overdue cycle id.
+- The monitor is observe-only: it reads `ApprenticeshipCycleStore`, computes age from `createdAt`,
+  and never closes, edits, or deletes cycle rows.
+- The server reuses `TokenLedgerPoller`'s existing cadence through an after-tick hook, avoiding a
+  new independent polling loop.
+
+Operational note: dedup is process-local. A server restart may allow one fresh attention item for an
+already-overdue cycle, but the default-off gate prevents surprise rollout noise and the route remains
+available as a pull surface.

package/upgrades/side-effects/circuit-breaker-reset-after.md

@@ -0,0 +1,51 @@
+# Side-Effects Review — Circuit-breaker "reset after Ns" parsing
+
+**Version / slug:** `circuit-breaker-reset-after`
+**Date:** `2026-06-03`
+**Author:** `echo`
+**Second-pass reviewer:** `not required`
+
+## Summary of the change
+
+Two regexes in `LlmCircuitBreaker.parseRetryAfterMs` change `\s+in\s+` → `\s+(?:in|after)\s+`,
+so a provider's "reset **after** N seconds/minutes" hint parses to `retryAfterMs` the same way
+"reset **in** N" already did. Adds three unit tests. No other code paths change.
+
+## Decision-point inventory
+
+None added. The change widens one existing best-effort parse; it introduces no new branch that
+decides anything. The parsed value flows into the same `onRateLimited` clamp that already existed.
+
+## 1. Over-block (false positive — a wrong/too-short window extracted)
+
+`parseRetryAfterMs` only runs inside `classifyRateLimit`, i.e. on a message ALREADY classified as a
+rate limit. The new alternation still requires the literal stem `reset`/`resets`/`try again`,
+immediately followed by `after`, a number, and a time unit — so it cannot fire on unrelated prose.
+Even if a genuine rate-limit message contained a misleading "reset after N" phrase, the extracted
+value is bounded twice: `classifyRateLimit` clamps to `[1s, 15min]`, and `onRateLimited` floors to
+`min(30s, openMs)`. The breaker therefore can never open for **less than 30s** or **more than 15min**
+regardless of what parses. The blast radius of a mis-parse is at most "recovers 30s vs the old 15min"
+— strictly safer than the bug being fixed.
+
+## 2. Under-block (false negative — a window still missed)
+
+Unchanged phrasings ("reset in Ns", "retry-after: N", "try again in N minutes") match exactly as
+before — the change is purely additive (an extra alternation), so no previously-parsed message
+stops parsing. Messages with no recognizable duration still return `undefined` and fall back to the
+flat default, identical to today.
+
+## 3. Interaction with #708 (gemini capacity policy)
+
+#708 handles the structured provider path (it retries within a single `evaluate` with the
+gemini-reported window). This fix is a layer up: it governs how long the GLOBAL breaker stays open
+when a gemini failure propagates as a classified rate-limit. The two compose — #708 keeps a single
+call resilient; this keeps a breaker trip proportionate. Neither overrides the other.
+
+## 4. Reversibility
+
+Pure code change, no migration, no persisted state, no config. Revert = revert the two regex edits.
+
+## Verdict
+
+Bounded, additive, no new decision surface. The only behavioral change is that gemini-style short
+resets now produce a proportionate (≥30s) breaker window instead of a flat 15-minute pause.