fix: add count-based fallback to archiveDecisions() (#626)

.squad/agents/capcom/history.md

@@ -64,3 +64,6 @@ Written full technical analysis to `.squad/identity/sdk-init-technical-analysis.
 - Template system in SDK (`templates/` directory) is well-structured, used correctly by `initSquad()`
 
 📌 **Team update (2026-03-11T01:25:00Z):** 5 SDK Init decisions merged to decisions.md: Phase-based quality improvement (3-phase approach), CastingEngine canonical casting, squad.config.ts as source of truth, Ralph always-included, implementation priority order. Full technical analysis informed Flight's unified PRD and EECOM's roadmap.
+
+📌 **Team update (2026-03-25T18:11Z):** CLI platform research complete — identified Copilot CLI 1.0.5–1.0.11 (8 releases in 10 days) contain three high-impact changes affecting Squad routing: monorepo instruction discovery (1.0.11), idle subagent hiding (1.0.8), subagentStart hook context injection (1.0.7). SDK version pinning doesn't prevent CLI runtime auto-updates. Recommendations: clean up template naming, upgrade to SDK 0.2.0 customize mode, file CLI issue. Report in decisions inbox.
+

.squad/agents/eecom/history.md

@@ -4,6 +4,21 @@
 
 ## Learnings
 
+### archiveDecisions() count-based fallback (#626) (2025-07-24)
+
+**Context:** `archiveDecisions()` in `packages/squad-cli/src/cli/core/nap.ts` silently returned `null` when all `###` entries were <30 days old (`old.length === 0`), even if the file was well over 20KB. Active projects generating many decisions per session could hit 145KB+ — 35K tokens burned per agent spawn.
+
+**Fix:** Added a count-based fallback after the age-based split. When `old.length === 0` and total file size exceeds `DECISION_THRESHOLD` (20KB), the fallback separates recent entries into dated vs undated, sorts dated by age (most recent first), keeps entries that fit under the threshold budget, and archives the rest. Undated entries are always preserved — they are foundational directives per Procedures' guidance.
+
+**Key design choices:**
+1. Undated entries (`daysAgo === null`) are never archived by the count-based fallback. They stay in `recent`.
+2. Budget calculation accounts for header + undated entries + kept dated entries to guarantee the result fits under 20KB.
+3. Entries are re-sorted into original document order after the split, so the output file preserves heading sequence.
+
+**Tests:** Added 4 adversarial tests — 50 all-today entries >20KB, mixed dated/undated preservation, under-threshold no-op, exact-threshold boundary case.
+
+**Pattern:** When a function has an early-return optimization (`if (old.length === 0) return null`), always consider whether the condition that triggered the function call (file size > threshold) can still be true when the early-return fires. If so, the early-return is a silent failure.
+
 ### Init scaffolding: casting dir + no-remote stderr (#579) (2025-07-18)
 
 **Context:** `squad init` in a fresh `git init` repo (no remote) printed `error: No such remote 'origin'` to stderr and `squad doctor` reported `casting/registry.json` missing. Two independent bugs in `packages/squad-sdk/src/config/init.ts`.
@@ -267,4 +282,6 @@ Executed 3 tasks across 2 waves: economy mode (#500, PR #504), node:sqlite fix (
 4. `personal.ts` - Refactored to reuse `ensurePersonalSquadDir()`.
 5. `resolution.test.ts` - Added 3 tests.
 
-**Pattern:** `resolveGlobalSquadPath()` returns the container; `ensurePersonalSquadDir()` creates the subdirectory the rest of the system looks for.
+**Pattern:** `resolveGlobalSquadPath()` returns the container; `ensurePersonalSquadDir()` creates the subdirectory the rest of the system looks for.
+📌 **Team update (2026-03-25T18:11Z):** Fixed #590 personal squad path regression — getPersonalSquadRoot() now uses canonical personal-squad/ subdirectory like esolvePersonalSquadDir() and nsurePersonalSquadDir(). Committed on squad/590-fix-personal-squad-root. FIDO found same bug in shell/index.ts → work passed to CONTROL for full sweep revision. Awaiting FIDO re-review.
+

.squad/agents/gnc/history.md

@@ -19,3 +19,7 @@ Reviewed and merged PR #474 (Node 22 ESM compatibility + bonus exports key fix).
 **Key learning:** ESM exports key and actual module structure must stay in sync. When adding new entry points, update both package.json (exports) and create the corresponding module file. Missing either step breaks consumers on Node 22+. Test matrix must include Node 22+ to catch these errors early.
 ### Dual-Layer ESM Fix (Issue #449)
 Upgraded from single-layer session.js patch to dual-layer approach: (1) Inject `exports` field into vscode-jsonrpc@8.2.1 package.json at postinstall — this is the canonical fix that resolves ALL subpath imports at once, matching vscode-jsonrpc v9.x. (2) Keep session.js `.js` extension patch as defense-in-depth. Added `squad doctor` detection for both layers (checks vscode-jsonrpc exports field and copilot-sdk session.js import syntax). Runtime Module._resolveFilename patch in cli-entry.ts remains as Layer 3 for npx cache hits where postinstall never runs.
+
+📌 **Team update (2026-03-25T18:11Z):** Routing regression research complete — root cause identified as combination of Squad code changes (v0.9.0 prompt saturation +33%, inlined workflows, missing 
+ame param in templates #577) AND Copilot CLI platform changes (CAPCOM report). Coordinator prompt grew 711→946 lines, diluting routing constraint. Workstream config replacement broke existing routing. Agent name fix exists on dev (#577) but not shipped. Recommendations: ship #577 immediately, move features to skills, optimize coordinator prompt, create regression tests. Report in decisions inbox.
+

.squad/agents/procedures/history.md

@@ -140,6 +140,8 @@ Also updated: examples section (showing `name` + `description` pairs), anti-patt
 
 📌 **Team update (2026-03-23T23:15Z):** Orchestration complete. Agent name display refactor shipped: spawn templates updated with mandatory `name` parameter across all 4 template variants. VOX and FIDO coordinated on parser extraction and cascading pattern strategies. All decisions merged to decisions.md. Canonical source: `.squad-templates/squad.agent.md` (all derived copies secondary).
 
+📌 **Team update (2026-03-25T18:11Z):** Model catalog updated to current platform offerings — removed 2 stale models (claude-opus-4.6-fast, gpt-5), added 5 new models (claude-sonnet-4.6, claude-opus-4.6-1m, gpt-5.4, gpt-5.3-codex, gpt-5.4-mini), bumped defaults (code: claude-sonnet-4.6, specialist: gpt-5.3-codex), restructured fallbacks. All 5 squad.agent.md template copies synchronized. Merged in #588.
+
 ### 2025-07: Model catalog refresh (#588)
 
 **Problem:** The valid models catalog, fallback chains, role-to-model mappings, and default model references in `squad.agent.md` were stale — missing `claude-sonnet-4.6`, `gpt-5.4`, `gpt-5.3-codex`, `gpt-5.4-mini`, `claude-opus-4.6-1m` and still referencing removed models `claude-opus-4.6-fast` and standalone `gpt-5`.