luiseiman · luiseiman · Jun 4, 2026 · Jun 4, 2026
diff --git a/.claude/hooks/session-start-process-overrides.sh b/.claude/hooks/session-start-process-overrides.sh
diff --git a/.claude/settings.json b/.claude/settings.json
@@ -49,11 +49,6 @@
           {
             "type": "command",
             "command": ".claude/hooks/session-startup.sh"
-          },
-          {
-            "type": "command",
-            "command": ".claude/hooks/session-start-process-overrides.sh",
-            "timeout": 5
           }
         ]
       }

diff --git a/.forge/audit/overrides.log b/.forge/audit/overrides.log
diff --git a/CLAUDE.md b/CLAUDE.md
@@ -73,7 +73,7 @@ Seven subagent definitions in `agents/`: researcher (read-only exploration), arc
 
 ### Audit System
 
-Two-dimension model (v4.x). **Dimension A — Native Health** (`score`, 0-10): 5 obligatory items (0-2) + 10 recommended (0-1), normalized as `obligatory*0.7 + recommended*0.3`. Security-critical items (settings.json, block-destructive hook) cap it at 6.0 if missing. Measures good use of native Claude Code (auto-memory as index, permission cascade, attribution, sandbox, deny rules). **Dimension B — dotforge Adoption** (`forge_adoption`, 0-5): behaviors/workflows/override-loop/domain-rules/sync-recency. **Informational — does NOT affect Native Health.** A native-first project scoring B=0 with A=10 is a desirable outcome (see `.claude/rules/domain/native-vs-dotforge-boundary.md`). `audit/checklist.md` + `audit/scoring.md` are the source of truth; registry in `registry/projects.yml` tracks both across managed projects. **Two scoring engines reimplement the checklist independently — `audit/score.sh` (bash, CI gate) and `scripts/audit_all.py` (Python, 12-project re-auditor). Any checklist change must update BOTH plus `audit.yml` and the docs (`README.md`, `docs/usage-guide.md`, `docs/guia-uso.md`); grep all consumers before planning the edit.**
+Two-dimension model (v4.x). **Dimension A — Native Health** (`score`, 0-10): 5 obligatory items (0-2) + 10 recommended (0-1), normalized as `obligatory*0.7 + recommended*0.3`. Security-critical items (settings.json, block-destructive hook) cap it at 6.0 if missing. Measures good use of native Claude Code (auto-memory as index, permission cascade, attribution, sandbox, deny rules). **Dimension B — dotforge Adoption** (`forge_adoption`, 0-4): behaviors/workflows/domain-rules/sync-recency. **Informational — does NOT affect Native Health.** A native-first project scoring B=0 with A=10 is a desirable outcome (see `.claude/rules/domain/native-vs-dotforge-boundary.md`). `audit/checklist.md` + `audit/scoring.md` are the source of truth; registry in `registry/projects.yml` tracks both across managed projects. **Two scoring engines reimplement the checklist independently — `audit/score.sh` (bash, CI gate) and `scripts/audit_all.py` (Python, 12-project re-auditor). Any checklist change must update BOTH plus `audit.yml` and the docs (`README.md`, `docs/usage-guide.md`, `docs/guia-uso.md`); grep all consumers before planning the edit.**
 
 ### Integrations
 

diff --git a/README.md b/README.md
@@ -11,15 +11,14 @@
 [![Version](https://img.shields.io/badge/version-4.0.0-blue)](VERSION)
 [![Last commit](https://img.shields.io/github/last-commit/luiseiman/dotforge)](https://github.com/luiseiman/dotforge/commits/main)
 
-**Behavior governance for [Claude Code](https://docs.anthropic.com/en/docs/claude-code).** Declare runtime policies on tool calls — "search before writing", "no destructive git", "verify before shipping" — and enforce them via compiled `PreToolUse` hooks that share a session-scoped state file. Escalates silently → nudge → warning → soft_block → hard_block, with a permanent override audit trail.
+**Behavior governance for [Claude Code](https://docs.anthropic.com/en/docs/claude-code).** Declare runtime policies on tool calls — "search before writing", "no destructive git", "verify before shipping" — and enforce them via compiled `PreToolUse` hooks that share a session-scoped state file. Escalates silently → nudge → warning → soft_block → hard_block.
 
 ```
 behaviors/no-destructive-git/behavior.yaml  # declarative policy
   ↓ (compile)
 .claude/hooks/generated/*.sh                # runtime enforcement
   ↓ (observe)
 .forge/runtime/state.json                   # counters, flags, per-session
-.forge/audit/overrides.log                  # override audit (git-tracked)
 ```
 
 Other tools stop at configuration. dotforge governs behavior — and keeps auditing, syncing, and evolving your `.claude/` setup across every repo you manage.
@@ -34,12 +33,11 @@ For people and teams managing more than one Claude Code project.
 
 ## v4.0 — what's new (2026-06-03)
 
-### Override capture loop closes practices↔behaviors (v4.0.0)
+### Audit two-dimension model + override loop retired (v4.0.x)
 
-- **`scripts/process-override-log.sh`** — bash script that processes `.forge/audit/overrides.log` and auto-creates `practices/inbox/auto-override-*.md` for behaviors overridden ≥3 times in 30 days. Idempotent. 10/10 tests green. Cost: 0 LLM calls, pure bash.
-- **`session-start-process-overrides.sh`** wired in `SessionStart` (template + self-hosting) — auto-captures frequent overrides as practices on every session start.
+- **Override capture loop — shipped in v4.0.0, retired after validation.** A portfolio scan found **0 overrides across all 12 projects in ~7 weeks** (production included; only 4/12 adopted behaviors at all). The auditable override trail + capture loop (`process-override-log.sh`, `overrides.log`, SessionStart wiring) were removed as dead weight. The graduated escalation engine (soft_block) — which IS exercised; the happy path is "verify", not "override" — stays. See [`.claude/rules/domain/native-vs-dotforge-boundary.md`](.claude/rules/domain/native-vs-dotforge-boundary.md).
 - **`scripts/migrate-v3-to-v4.sh`** — safe migration script with mandatory `--dry-run`, atomic backup, `--rollback`. See [`docs/v4/MIGRATION-V3-TO-V4.md`](docs/v4/MIGRATION-V3-TO-V4.md).
-- **Audit two-dimension model** — **Native Health** (0-10: native Claude Code usage + security) + **dotforge Adoption** (0-5: informational, does not affect the score). Behaviors / workflows / override-loop moved to the non-penalizing Adoption dimension — native-first projects no longer lose points for skipping dotforge machinery. New Native-Health items: auto-memory hygiene, permission cascade, attribution.
+- **Audit two-dimension model** — **Native Health** (0-10: native Claude Code usage + security) + **dotforge Adoption** (0-4: informational, does not affect the score). Behaviors / workflows moved to the non-penalizing Adoption dimension — native-first projects no longer lose points for skipping dotforge machinery. New Native-Health items: auto-memory hygiene, permission cascade, attribution.
 - **`domain/workflow-economics.md`** (new domain rule) — documents v4 PoC cost-quality findings. Decision matrix: when workflow vs skill. Token economy principles. **TL;DR: workflows are 4-25x more expensive than bash skills for recurring work — use only as on-demand escalation, not as default refactor.**
 - **`workflows/watch.js`** ships as REFERENCE implementation, NOT promoted to `/forge watch` default. The bash skill remains the production tool.
 
@@ -300,7 +298,7 @@ Orchestration follows a decision tree: researcher → architect → implementer
 `/forge audit` scores your project's Claude Code configuration on a 10-point scale:
 
 - **5 obligatory items** (scored 0-2): CLAUDE.md, settings.json, rules with globs, block-destructive hook, build/test commands
-- **12 recommended items** (scored 0-1): CLAUDE_ERRORS.md, lint hook, custom commands, memory, agents + orchestration rule, .gitignore secrets, prompt injection scan, auto-mode safety, v3 behaviors enforcement, OS-level sandboxing, **v4 workflow availability**, **v4 override capture loop active**
+- **12 recommended items** (scored 0-1): CLAUDE_ERRORS.md, lint hook, custom commands, memory, agents + orchestration rule, .gitignore secrets, prompt injection scan, auto-mode safety, v3 behaviors enforcement, OS-level sandboxing, **v4 workflow availability**, domain rules
 - **Project tier**: simple/standard/complex adjusts scoring expectations
 - **Security cap**: missing settings.json or block-destructive hook caps score at 6.0
 

diff --git a/audit/checklist.md b/audit/checklist.md
@@ -3,7 +3,7 @@
 El audit tiene **dos dimensiones independientes**:
 
 - **A — Salud Nativa** (score 0-10): ¿el proyecto usa bien Claude Code nativo + seguridad? Es el score que importa para cualquier proyecto, use o no la maquinaria dotforge.
-- **B — Adopción dotforge** (informativo 0-5): ¿cuánto adoptó la gobernanza dotforge? **NO penaliza** la Salud Nativa. Un proyecto native-first puro saca 0/5 acá sin perder un punto en A.
+- **B — Adopción dotforge** (informativo 0-4): ¿cuánto adoptó la gobernanza dotforge? **NO penaliza** la Salud Nativa. Un proyecto native-first puro saca 0/4 acá sin perder un punto en A.
 
 ---
 
@@ -100,9 +100,9 @@ El audit tiene **dos dimensiones independientes**:
 
 ---
 
-# Dimensión B — Adopción dotforge (informativo, 0-5)
+# Dimensión B — Adopción dotforge (informativo, 0-4)
 
-**No afecta el score de Salud Nativa.** Mide cuánto adoptó el proyecto la maquinaria de gobernanza dotforge. Reportar como `Adopción: N/5` con label (0=None, 1-2=Partial, 3-4=Most, 5=Full). Sirve para decidir propagación, no para juzgar calidad.
+**No afecta el score de Salud Nativa.** Mide cuánto adoptó el proyecto la maquinaria de gobernanza dotforge. Reportar como `Adopción: N/4` con label (0=None, 1-2=Partial, 3=Most, 4=Full). Sirve para decidir propagación, no para juzgar calidad.
 
 ### B1. Behaviors v3 compilados y wired
 - 0: Sin behaviors enforced — declaración en `behaviors/index.yaml` sola NO cuenta
@@ -116,19 +116,13 @@ El audit tiene **dos dimensiones independientes**:
 
 **Verificación:** `grep -q "export const meta" workflows/*.js`. Señal de gobernanza, no de calidad — los bash skills siguen siendo el workhorse. Ver `docs/v4/SPEC.md`.
 
-### B3. Override capture loop activo (v4)
-- 0: `.forge/audit/overrides.log` no rastreado O `session-start-process-overrides.sh` no wired
-- 1: Ambos presentes: log existe Y el hook está en `.claude/settings.json` SessionStart
-
-**Verificación:** `test -f .forge/audit/overrides.log && grep -q "session-start-process-overrides.sh" .claude/settings.json`. Solo significativo si hay behaviors activos. Ver `scripts/process-override-log.sh`.
-
-### B4. Domain rules
+### B3. Domain rules
 - 0: No hay `.claude/rules/domain/`
 - 1: Al menos un domain rule presente y fresco (`last_verified` <90 días)
 
 **Verificación:** Contar archivos en `.claude/rules/domain/`. Reportar cuántos están stale (>90 días). Si hay lógica de negocio pero no domain rules, sugerir `/forge domain extract`.
 
-### B5. Sync recency
+### B4. Sync recency
 - 0: `dotforge_version` del proyecto desfasado respecto a `VERSION` por ≥1 minor, o desconocido
 - 1: Proyecto sincronizado a la versión actual de dotforge (`dotforge_version` == `VERSION`)
 

diff --git a/audit/score.sh b/audit/score.sh
@@ -7,7 +7,7 @@
 # Two-dimension model (v4.x — see audit/checklist.md + audit/scoring.md):
 #   Dimension A — Native Health: 5 obligatory (0-2) + 10 recommended (0-1).
 #                 score = obl*0.7 + rec*0.3, security cap 6.0. The CI gate.
-#   Dimension B — dotforge Adoption: 5 items (0-1). Informational, 0-5.
+#   Dimension B — dotforge Adoption: 4 items (0-1). Informational, 0-4.
 #                 Does NOT affect Native Health.
 # Semantic checks (CLAUDE.md quality, rule content) are approximated with heuristics.
 # Score is indicative — /forge audit provides authoritative semantic evaluation.
@@ -44,8 +44,8 @@ cd "$PROJECT_DIR"
 s1=0; n1=""; s2=0; n2=""; s3=0; n3=""; s4=0; n4=""; s5=0; n5=""
 s6=0; n6=""; s7=0; n7=""; s8=0; n8=""; s9=0; n9=""; s10=0; n10=""
 s11=0; n11=""; s12=0; n12=""; s13=0; n13=""; s14=0; n14=""; s15=0; n15=""
-# --- Dimension B: scores (b1..b5) and notes (m1..m5) ---
-b1=0; m1=""; b2=0; m2=""; b3=0; m3=""; b4=0; m4=""; b5=0; m5=""
+# --- Dimension B: scores (b1..b4) and notes (m1..m4) ---
+b1=0; m1=""; b2=0; m2=""; b3=0; m3=""; b4=0; m4=""
 
 # ─────────────────────────────────────────────────────────────────────────────
 # DIMENSION A — OBLIGATORIO (each 0-2)
@@ -320,24 +320,16 @@ else
   b2=0; m2="No workflows/ with valid meta block"
 fi
 
-# B3. Override capture loop active (v4)
-if [[ -f ".forge/audit/overrides.log" ]] && [[ -f "$SETTINGS" ]] \
-   && grep -q "session-start-process-overrides.sh" "$SETTINGS" 2>/dev/null; then
-  b3=1; m3="overrides.log present and hook wired in SessionStart"
-else
-  b3=0; m3="Override loop not wired (log + SessionStart hook required)"
-fi
-
-# B4. Domain rules present
+# B3. Domain rules present
 DOM=$(ls .claude/rules/domain/*.md 2>/dev/null | wc -l | tr -d ' ')
 if [[ "${DOM:-0}" -gt 0 ]]; then
-  b4=1; m4="${DOM} domain rule(s) (freshness checked semantically by /forge audit)"
+  b3=1; m3="${DOM} domain rule(s) (freshness checked semantically by /forge audit)"
 else
-  b4=0; m4="No domain rules in .claude/rules/domain/"
+  b3=0; m3="No domain rules in .claude/rules/domain/"
 fi
 
-# B5. Sync recency — not mechanically determinable standalone (needs registry)
-b5=0; m5="Sync recency indeterminate standalone — resolved by /forge audit via registry"
+# B4. Sync recency — not mechanically determinable standalone (needs registry)
+b4=0; m4="Sync recency indeterminate standalone — resolved by /forge audit via registry"
 
 # ─────────────────────────────────────────────────────────────────────────────
 # Calculate scores
@@ -352,10 +344,10 @@ if [[ $s2 -eq 0 || $s4 -eq 0 ]]; then
   NATIVE_HEALTH=$(awk "BEGIN { v=${NATIVE_HEALTH}; printf \"%.2f\", (v > 6.0 ? 6.0 : v) }")
 fi
 
-FORGE_ADOPTION=$((b1 + b2 + b3 + b4 + b5))
+FORGE_ADOPTION=$((b1 + b2 + b3 + b4))
 if   [[ $FORGE_ADOPTION -eq 0 ]]; then ADOPTION_LABEL="None"
 elif [[ $FORGE_ADOPTION -le 2 ]]; then ADOPTION_LABEL="Partial"
-elif [[ $FORGE_ADOPTION -le 4 ]]; then ADOPTION_LABEL="Most"
+elif [[ $FORGE_ADOPTION -eq 3 ]]; then ADOPTION_LABEL="Most"
 else                                   ADOPTION_LABEL="Full"
 fi
 
@@ -409,9 +401,8 @@ data = {
   "adoption_items": {
     "B1_behaviors":        {"score": ${b1}, "note": "$(_san "$m1")"},
     "B2_workflows":        {"score": ${b2}, "note": "$(_san "$m2")"},
-    "B3_override_loop":    {"score": ${b3}, "note": "$(_san "$m3")"},
-    "B4_domain_rules":     {"score": ${b4}, "note": "$(_san "$m4")"},
-    "B5_sync_recency":     {"score": ${b5}, "note": "$(_san "$m5")"}
+    "B3_domain_rules":     {"score": ${b3}, "note": "$(_san "$m3")"},
+    "B4_sync_recency":     {"score": ${b4}, "note": "$(_san "$m4")"}
   }
 }
 print(json.dumps(data, indent=2))
@@ -421,7 +412,7 @@ else
   $SECURITY_CAP && CAP_NOTE="  ⚠ security cap applied (settings.json or block-destructive missing)"
   echo "═══ AUDIT SCORE: $(basename "$PROJECT_DIR") ═══"
   echo "Native Health: ${NATIVE_HEALTH}/10  (${LEVEL})${CAP_NOTE}"
-  echo "dotforge Adoption: ${FORGE_ADOPTION}/5  (${ADOPTION_LABEL})  [informational — does not affect Native Health]"
+  echo "dotforge Adoption: ${FORGE_ADOPTION}/4  (${ADOPTION_LABEL})  [informational — does not affect Native Health]"
   echo ""
   echo "═ DIMENSION A — NATIVE HEALTH ═"
   echo "── OBLIGATORIO (${SCORE_OBL}/10) ──"
@@ -446,9 +437,8 @@ else
   echo "═ DIMENSION B — DOTFORGE ADOPTION (informational) ═"
   printf "  [%s] B1. v3 behaviors         %s\n" "$b1" "$m1"
   printf "  [%s] B2. Workflow available   %s\n" "$b2" "$m2"
-  printf "  [%s] B3. Override loop        %s\n" "$b3" "$m3"
-  printf "  [%s] B4. Domain rules         %s\n" "$b4" "$m4"
-  printf "  [%s] B5. Sync recency         %s\n" "$b5" "$m5"
+  printf "  [%s] B3. Domain rules         %s\n" "$b3" "$m3"
+  printf "  [%s] B4. Sync recency         %s\n" "$b4" "$m4"
 fi
 
 # CI threshold gate (on Native Health)

diff --git a/audit/scoring.md b/audit/scoring.md
@@ -1,6 +1,6 @@
 # Scoring de Auditoría
 
-El audit produce **dos números independientes**: `native_health` (0-10, el score principal) y `forge_adoption` (0-5, informativo).
+El audit produce **dos números independientes**: `native_health` (0-10, el score principal) y `forge_adoption` (0-4, informativo).
 
 ## Dimensión A — Salud Nativa (score principal)
 
@@ -34,7 +34,7 @@ Si alguno de estos items es **0**, `native_health` tiene un cap máximo de **6.0
 ## Dimensión B — Adopción dotforge (informativo)
 
 ```
-forge_adoption = sum(items B1-B5)   # 0-5
+forge_adoption = sum(items B1-B4)   # 0-4
 ```
 
 **No entra en `native_health` ni lo modifica.** Es un indicador de cuánta gobernanza dotforge adoptó el proyecto.
@@ -43,8 +43,8 @@ forge_adoption = sum(items B1-B5)   # 0-5
 |----|-------|---------|
 | 0   | None    | Native-first puro. Válido y sin penalización. |
 | 1-2 | Partial | Adopción parcial de la maquinaria. |
-| 3-4 | Most    | Adopción amplia. |
-| 5   | Full    | Gobernanza dotforge completa. |
+| 3   | Most    | Adopción amplia. |
+| 4   | Full    | Gobernanza dotforge completa. |
 
 Un `forge_adoption: 0` con `native_health: 10` es un resultado **excelente y deseable** bajo el principio native-first (ver `.claude/rules/domain/native-vs-dotforge-boundary.md`). No recomendar adoptar maquinaria dotforge solo para subir B.
 

diff --git a/behaviors/search-first/tests/scenario_override_reinvocation.sh b/behaviors/search-first/tests/scenario_override_reinvocation.sh
@@ -1,7 +1,7 @@
 #!/usr/bin/env bash
 # Override reinvocation: escalate to soft_block, then the exact same tool_input
 # comes back → detected as override, passes silently, records audit in three
-# places (state.overrides, .forge/audit/overrides.log, pending_block cleared).
+# places (state.overrides, pending_block cleared).
 set -u
 . "$(dirname "$0")/_scenario_helpers.sh"
 trap scenario_cleanup EXIT
@@ -48,10 +48,6 @@ ov_counter=$(jq -r --arg sid "$SCENARIO_SESSION_ID" \
     '.sessions[$sid].behaviors["search-first"].overrides[0].counter_at_override' "$FORGE_STATE_FILE")
 assert_eq "5" "$ov_counter" "override captured counter=5" || exit 1
 
-# .forge/audit/overrides.log has 1 line
-audit_lines=$(wc -l < "${FORGE_ROOT}/audit/overrides.log" | tr -d ' ')
-assert_eq "1" "$audit_lines" "audit log has one override entry" || exit 1
-
 # pending_block cleared
 pending_after=$(jq -r --arg sid "$SCENARIO_SESSION_ID" \
     '.sessions[$sid].behaviors["search-first"].pending_block // "null"' "$FORGE_STATE_FILE")