diff --git a/CHANGELOG.md b/CHANGELOG.md
index 305478b..b0b11f6 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,16 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ## [Unreleased]
 
+## [3.32.1] - 2026-03-08
+
+### Added
+
+- **`auto-rename-session.sh` hook template** (`examples/hooks/bash/auto-rename-session.sh`) — hook SessionEnd qui génère automatiquement un titre descriptif pour chaque session. Lit le JSONL de session directement, extrait les 3 premiers messages utilisateur, appelle `claude -p --model claude-haiku-4-5-20251001` pour générer un titre 4-6 mots (format `verb + subject`), fallback sur le premier message nettoyé si Haiku indisponible. Met à jour le slug dans le JSONL natif (pour `/resume`) et dans `sessions-index.jsonl`. Output via `/dev/tty` pour bypasser le parsing JSON de Claude Code.
+
+### Documentation
+
+- **Section "Session Auto-Rename" mise à jour** (`guide/ultimate-guide.md`) — présente désormais deux approches complémentaires : Approche A (instruction CLAUDE.md, renommage mid-session via `/rename`, zéro tooling) et Approche B (hook SessionEnd, titre généré par Haiku en post-session, lecture directe du JSONL). Suppression du paragraphe "Why not a hook?" qui était incorrect depuis l'introduction de l'accès aux données de session via JSONL.
+
 ## [3.32.0] - 2026-03-06
 
 ### Added
diff --git a/examples/hooks/bash/auto-rename-session.sh b/examples/hooks/bash/auto-rename-session.sh
new file mode 100755
index 0000000..c4df0b7
--- /dev/null
+++ b/examples/hooks/bash/auto-rename-session.sh
@@ -0,0 +1,252 @@
+#!/bin/bash
+# .claude/hooks/auto-rename-session.sh
+# Event: SessionEnd
+# Automatically generates a descriptive title for each session when it ends.
+#
+# Strategy:
+#   1. Reads the session JSONL from ~/.claude/projects/ (via session_id from stdin or env)
+#   2. Extracts the first 3 user messages as context
+#   3. Calls claude-haiku-4-5 to generate a 4-6 word title (verb + subject)
+#   4. Falls back to a sanitized version of the first message if Haiku fails
+#   5. Updates sessions-index.jsonl (custom index) and the JSONL slug (for /resume)
+#
+# Requirements:
+#   - python3 (standard on macOS/Linux)
+#   - claude CLI on PATH (for AI title generation; fallback works without it)
+#
+# Wire up in .claude/settings.json:
+#   {
+#     "hooks": {
+#       "SessionEnd": [
+#         {
+#           "matcher": "",
+#           "hooks": [{ "type": "command", "command": "~/.claude/hooks/auto-rename-session.sh" }]
+#         }
+#       ]
+#     }
+#   }
+#
+# Configuration:
+#   SESSION_AUTORENAME=0   Disable this hook for a specific session
+#
+# Note: Output goes to /dev/tty (or /dev/stderr as fallback) to avoid interfering
+# with Claude Code's JSON parsing of hook stdout.
+
+set -uo pipefail
+
+# Route output to tty so it doesn't pollute Claude Code's JSON parsing of stdout
+if (echo -n "" > /dev/tty) 2>/dev/null; then
+    OUT="/dev/tty"
+else
+    OUT="/dev/stderr"
+fi
+
+# Opt-out mechanism: set SESSION_AUTORENAME=0 in your shell or .env to skip
+[[ "${SESSION_AUTORENAME:-1}" == "0" ]] && exit 0
+
+# ---------------------------------------------------------------------------
+# Step 1: Resolve session_id and working directory
+# ---------------------------------------------------------------------------
+# Claude Code passes a JSON payload on stdin with session_id and cwd.
+# If stdin was already consumed by a preceding hook, fall back to env vars.
+
+INPUT=$(cat 2>/dev/null || true)
+
+SESSION_ID=""
+CWD=""
+
+if [[ -n "$INPUT" ]]; then
+    SESSION_ID=$(echo "$INPUT" | python3 -c "
+import sys, json
+try:
+    d = json.load(sys.stdin)
+    print(d.get('session_id', ''))
+except: pass
+" 2>/dev/null || true)
+
+    CWD=$(echo "$INPUT" | python3 -c "
+import sys, json
+try:
+    d = json.load(sys.stdin)
+    print(d.get('cwd', ''))
+except: pass
+" 2>/dev/null || true)
+fi
+
+# Fall back to environment variables if stdin parsing yielded nothing
+[[ -z "$SESSION_ID" ]] && SESSION_ID="${CLAUDE_SESSION_ID:-}"
+[[ -z "$CWD" ]] && CWD="${CLAUDE_PROJECT_DIR:-$(pwd)}"
+[[ -z "$CWD" ]] && CWD="$(pwd)"
+
+# ---------------------------------------------------------------------------
+# Step 2: Locate the session JSONL file
+# ---------------------------------------------------------------------------
+# Claude Code stores sessions under ~/.claude/projects/<url-encoded-cwd>/<session_id>.jsonl
+
+PROJECT_SLUG=$(echo "$CWD" | sed 's|/|-|g')
+PROJECTS_DIR="$HOME/.claude/projects/$PROJECT_SLUG"
+
+JSONL_FILE=""
+
+if [[ -n "$SESSION_ID" && -f "$PROJECTS_DIR/$SESSION_ID.jsonl" ]]; then
+    JSONL_FILE="$PROJECTS_DIR/$SESSION_ID.jsonl"
+elif [[ -d "$PROJECTS_DIR" ]]; then
+    # No explicit session_id: pick the most recently modified JSONL (excluding agent files)
+    JSONL_FILE=$(ls -t "$PROJECTS_DIR"/*.jsonl 2>/dev/null | grep -v '/agent-' | head -1 || true)
+    [[ -n "$JSONL_FILE" ]] && SESSION_ID=$(basename "$JSONL_FILE" .jsonl)
+fi
+
+[[ -z "$JSONL_FILE" || ! -f "$JSONL_FILE" ]] && exit 0
+
+# Read the current slug so we can detect whether it changed later
+CURRENT_SLUG=$(python3 -c "
+import json, sys
+with open(sys.argv[1]) as f:
+    for line in f:
+        try:
+            obj = json.loads(line)
+            s = obj.get('slug', '')
+            if s: print(s); break
+        except: pass
+" "$JSONL_FILE" 2>/dev/null || true)
+
+# ---------------------------------------------------------------------------
+# Step 3: Extract the first 3 user messages as context
+# ---------------------------------------------------------------------------
+# Skips tool result blobs (<...), slash commands (/...), and messages longer
+# than 250 characters to keep the prompt tight.
+
+CONTEXT=$(python3 -c "
+import json, sys
+
+jsonl_file = sys.argv[1]
+messages = []
+
+with open(jsonl_file) as f:
+    for line in f:
+        try:
+            obj = json.loads(line)
+            if obj.get('type') != 'user':
+                continue
+            content = obj.get('message', {}).get('content', '')
+            if not isinstance(content, str):
+                continue
+            content = content.strip()
+            if content.startswith('<') or content.startswith('/'):
+                continue
+            messages.append(content[:250])
+            if len(messages) >= 3:
+                break
+        except:
+            continue
+
+print('\n---\n'.join(messages))
+" "$JSONL_FILE" 2>/dev/null || true)
+
+[[ -z "$CONTEXT" ]] && exit 0
+
+FIRST_MSG=$(echo "$CONTEXT" | head -1 | head -c 60)
+
+# ---------------------------------------------------------------------------
+# Step 4: Generate a title via Haiku (with plaintext fallback)
+# ---------------------------------------------------------------------------
+
+TITLE=""
+TITLE_SOURCE="fallback"
+
+PROMPT="Generate a 4-6 word lowercase English title summarizing this Claude Code session. Format: verb + subject (e.g. 'fix auth bug postgres', 'add stripe webhook handler', 'eval session labels tool'). Reply with ONLY the title, nothing else.
+
+Session messages:
+$CONTEXT"
+
+if command -v claude &>/dev/null; then
+    # CLAUDECODE is unset to prevent the subprocess from inheriting hook context,
+    # which could cause unexpected behaviour or infinite loops.
+    TITLE=$(printf '%s' "$PROMPT" \
+        | env -u CLAUDECODE timeout 12 claude -p --model claude-haiku-4-5-20251001 2>/dev/null \
+        | head -1 \
+        | tr '[:upper:]' '[:lower:]' \
+        | sed 's/[^a-z0-9 ]//g' \
+        | sed 's/  */ /g' \
+        | sed 's/^ //;s/ $//' \
+        | head -c 60 \
+        || true)
+    [[ -n "$TITLE" ]] && TITLE_SOURCE="haiku"
+fi
+
+# Fallback: sanitize the first user message into a slug-friendly string
+if [[ -z "$TITLE" ]]; then
+    TITLE=$(echo "$FIRST_MSG" \
+        | tr '[:upper:]' '[:lower:]' \
+        | sed 's/[^a-z0-9 ]/ /g' \
+        | sed 's/  */ /g' \
+        | sed 's/^ //;s/ $//' \
+        | head -c 55 \
+        || true)
+fi
+
+[[ -z "$TITLE" ]] && exit 0
+
+SLUG=$(echo "$TITLE" | sed 's/ /-/g' | sed 's/[^a-z0-9-]//g')
+
+# ---------------------------------------------------------------------------
+# Step 5: Persist the title
+# ---------------------------------------------------------------------------
+
+# 5a. Update sessions-index.jsonl (used by custom session browsers / tooling)
+python3 -c "
+import json, os, sys
+
+index_path = os.path.expanduser('~/.claude/sessions-index.jsonl')
+session_id = sys.argv[1]
+title = sys.argv[2]
+
+try:
+    with open(index_path) as f:
+        entries = [json.loads(line) for line in f if line.strip()]
+    for entry in entries:
+        if entry['id'] == session_id:
+            entry['context'] = title
+            break
+    with open(index_path, 'w') as f:
+        for entry in entries:
+            f.write(json.dumps(entry) + '\n')
+except:
+    pass
+" "$SESSION_ID" "$TITLE" 2>/dev/null || true
+
+# 5b. Update the slug field inside the JSONL so /resume shows the new title
+if [[ -n "$CURRENT_SLUG" && -n "$SLUG" && "$CURRENT_SLUG" != "$SLUG" ]]; then
+    TMPFILE=$(mktemp)
+    python3 -c "
+import json, sys
+
+jsonl_file, old_slug, new_slug = sys.argv[1], sys.argv[2], sys.argv[3]
+
+with open(jsonl_file) as f:
+    lines = f.readlines()
+
+out = []
+for line in lines:
+    try:
+        obj = json.loads(line)
+        if obj.get('slug') == old_slug:
+            obj['slug'] = new_slug
+        out.append(json.dumps(obj, separators=(',', ':')))
+    except:
+        out.append(line.rstrip())
+
+print('\n'.join(out))
+" "$JSONL_FILE" "$CURRENT_SLUG" "$SLUG" > "$TMPFILE" 2>/dev/null
+
+    [[ -s "$TMPFILE" ]] && mv "$TMPFILE" "$JSONL_FILE" || rm -f "$TMPFILE"
+fi
+
+# ---------------------------------------------------------------------------
+# Done
+# ---------------------------------------------------------------------------
+echo "" > "$OUT"
+echo "Session renamed: \"$TITLE\" ($TITLE_SOURCE)" > "$OUT"
+echo "" > "$OUT"
+
+exit 0
diff --git a/guide/ultimate-guide.md b/guide/ultimate-guide.md
index 0c5d375..6f6fcc4 100644
--- a/guide/ultimate-guide.md
+++ b/guide/ultimate-guide.md
@@ -858,9 +858,13 @@ Claude: [Resumes with Serena's persistent project understanding]
 
 ### Session Auto-Rename
 
-When running multiple Claude Code sessions in parallel (split terminals, WebStorm tabs, parallel workstreams), the `/resume` picker shows sessions by timestamp or truncated first prompt — impossible to distinguish.
+When running multiple Claude Code sessions in parallel (split terminals, WebStorm tabs, parallel workstreams), the `/resume` picker shows sessions by timestamp or truncated first prompt — impossible to distinguish at a glance.
 
-**Solution**: A behavioral instruction in `~/.claude/CLAUDE.md` that makes Claude rename sessions automatically after 2-3 exchanges, with no tooling required.
+Two complementary approaches solve this. Use one or both together.
+
+#### Approach A: CLAUDE.md behavioral instruction (mid-session)
+
+A behavioral instruction in `~/.claude/CLAUDE.md` makes Claude call `/rename` automatically after 2-3 exchanges. No tooling required, works across all IDEs and terminals.
 
 ```markdown
 # Session Naming (auto-rename)
@@ -887,13 +891,45 @@ When running multiple Claude Code sessions in parallel (split terminals, WebStor
 - Do NOT ask for confirmation on early rename (just do it)
 ```
 
-**Why not a hook?** The `Stop` event hook has no access to conversation context — it can't infer a meaningful title. The behavioral instruction approach costs zero tooling and works across all IDEs and terminals.
+This works well during active sessions but depends on Claude following the instruction consistently.
 
-**Limitation**: Terminal tab names (WebStorm, iTerm2) are not affected. JetBrains filters ANSI escape sequences. The Claude session is renamed, not the OS tab.
+#### Approach B: SessionEnd hook (automatic, AI-generated)
 
-After a session, the `/resume` picker shows `"fix auth middleware"` instead of `"2026-03-04T14:23..."`.
+A `SessionEnd` hook reads the session's JSONL file directly from `~/.claude/projects/`, extracts the first few user messages as context, and calls `claude -p --model claude-haiku-4-5-20251001` to generate a 4-6 word descriptive title. If Haiku is unavailable, it falls back to a sanitized version of the first message.
+
+The hook updates both `sessions-index.jsonl` (for custom session browsers) and the slug field in the JSONL file (for native `/resume` compatibility).
+
+```json
+// .claude/settings.json
+{
+  "hooks": {
+    "SessionEnd": [
+      {
+        "matcher": "",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "~/.claude/hooks/auto-rename-session.sh"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+Requirements: `claude` CLI on PATH, `python3` for JSON parsing. Set `SESSION_AUTORENAME=0` to disable for a specific session.
+
+After the session ends, the `/resume` picker shows `"fix auth middleware"` instead of `"2026-03-04T14:23..."`.
+
+#### Using both together
+
+The two approaches handle different moments in a session's lifecycle. Approach A renames early so the session is identifiable while it's still running. Approach B renames at the end with a title that reflects the full session scope, potentially overwriting the mid-session name with something more accurate.
+
+**Limitation (both approaches)**: Terminal tab names in WebStorm and iTerm2 are not affected. JetBrains filters ANSI escape sequences. The Claude session is renamed, not the OS tab.
 
 > See full template: [examples/claude-md/session-naming.md](../examples/claude-md/session-naming.md)
+> See hook template: [examples/hooks/bash/auto-rename-session.sh](../examples/hooks/bash/auto-rename-session.sh)
 
 ## 1.4 Permission Modes