Fix heartbeat path resolution and add model selection

Heartbeat prompt now references $NANOCLAW_GROUP_DIR/HEARTBEAT.md explicitly
so the agent can find the checklist. Fixed stale nanoclaw log paths in
heartbeat skill. Added GHOSTCLAW_MODEL env var for model selection
(sonnet/opus/haiku), wired through agent runner to Claude SDK. Setup
flow now asks which model and defaults to Sonnet. Phase 8 extras now
include update-check and use multiSelect.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: ziggy

.claude/skills/add-heartbeat/SKILL.md

@@ -26,7 +26,7 @@ Run these checks periodically. Only message the user if something needs attentio
 
 ## Checks
 
-- [ ] **NanoClaw errors**: `tail -20 ~/nanoclaw/logs/nanoclaw.error.log` — alert if non-empty in the last hour
+- [ ] **GhostClaw errors**: `tail -20 ~/nanoclaw/logs/ghostclaw.error.log` — alert if non-empty in the last hour
 - [ ] **Disk space**: `df -h /` — alert if usage >80%
 - [ ] **Unread emails**: Check Gmail for unread emails in the last hour — summarise if any look urgent
 
@@ -56,7 +56,7 @@ mkdir -p "$NANOCLAW_DATA_DIR/ipc/main/tasks"
 cat > "$NANOCLAW_DATA_DIR/ipc/main/tasks/heartbeat_$(date +%s).json" << 'EOF'
 {
   "type": "schedule_task",
-  "prompt": "Read HEARTBEAT.md and run each check listed. Only message the user if something needs attention. If everything is fine, respond with <internal>All checks passed</internal> and nothing else.",
+  "prompt": "Read $NANOCLAW_GROUP_DIR/HEARTBEAT.md (or ./HEARTBEAT.md in the current directory) and run each check listed. Only message the user if something needs attention. If everything is fine, respond with <internal>All checks passed</internal> and nothing else.",
   "schedule_type": "cron",
   "schedule_value": "*/30 * * * *",
   "context_mode": "isolated"
@@ -71,7 +71,7 @@ Adjust the cron value based on the user's preference.
 Wait for the next cron trigger (or manually create a test by adding a failing check to HEARTBEAT.md), then check logs:
 
 ```bash
-grep -i "heartbeat\|All checks passed" ~/nanoclaw/logs/nanoclaw.log | tail -5
+grep -i "heartbeat\|All checks passed" ~/nanoclaw/logs/ghostclaw.log | tail -5
 ```
 
 ## Customisation

.claude/skills/setup-ghostclaw/SKILL.md

@@ -39,6 +39,22 @@ CLAUDE_CODE_OAUTH_TOKEN=<token>
 ANTHROPIC_API_KEY=<key>
 ```
 
+### Model selection
+
+AskUserQuestion: Which model should the bot use?
+- Sonnet (Recommended) — fast, capable, cost-effective for most tasks
+- Opus — most capable, slower, higher cost
+- Haiku — fastest, cheapest, good for simple tasks
+
+Add to `.env` based on choice:
+```
+GHOSTCLAW_MODEL=claude-sonnet-4-6    # Sonnet (recommended)
+# GHOSTCLAW_MODEL=claude-opus-4-6    # Opus
+# GHOSTCLAW_MODEL=claude-haiku-4-5-20251001  # Haiku
+```
+
+If they skip or aren't sure, default to Sonnet. Tell them: "You can change this any time by editing `GHOSTCLAW_MODEL` in `.env` and restarting."
+
 ### Bot name
 
 AskUserQuestion: What should the bot be called?
@@ -237,9 +253,12 @@ If working: "You're set. Talk to BOTNAME — it knows who you are."
 
 ## Phase 8: Optional extras
 
-After core setup, offer:
+After core setup, offer optional extras. Use AskUserQuestion with multiSelect:
 
-- `/add-heartbeat` — periodic monitoring checks
-- `/add-morning-briefing` — scheduled briefings
-- `/add-gmail-agent` — email access for the bot
+- `/add-heartbeat` — periodic monitoring checks (recommended)
+- `/add-update-check` — weekly check for GhostClaw updates (recommended)
+- `/add-morning-briefing` — scheduled daily/weekly briefings
+- `/add-gmail-agent` — email access for the bot (needs Google Cloud OAuth)
 - `/add-voice` — voice note transcription (needs OpenAI key)
+
+Run each selected skill in sequence. If none selected, skip — they can always add these later.

.env.example

@@ -15,5 +15,10 @@ TELEGRAM_ONLY=true
 # Voice transcription (optional — needs OpenAI billing)
 # OPENAI_API_KEY=
 
+# Model (optional — defaults to your subscription's best model)
+# Options: claude-sonnet-4-6, claude-opus-4-6, claude-haiku-4-5-20251001
+# Sonnet recommended for most use cases (fast, cheap, capable)
+# GHOSTCLAW_MODEL=claude-sonnet-4-6
+
 # Gmail integration (optional — needs OAuth setup)
 # GMAIL_MCP_ENABLED=1

README.md

@@ -166,6 +166,7 @@ All config lives in `.env`. See `.env.example` for all options.
 | `ASSISTANT_NAME` | Yes | Bot name (trigger word in groups) |
 | `TELEGRAM_BOT_TOKEN` | Recommended | From @BotFather |
 | `TELEGRAM_ONLY` | No | Set `true` to skip WhatsApp |
+| `GHOSTCLAW_MODEL` | No | Model ID (default: subscription best). Options: `claude-sonnet-4-6`, `claude-opus-4-6`, `claude-haiku-4-5-20251001` |
 | `OPENAI_API_KEY` | No | For voice transcription |
 | `GMAIL_MCP_ENABLED` | No | Set `1` for Gmail integration |
 

container/agent-runner/src/index.ts

@@ -422,6 +422,7 @@ async function runQuery(
     prompt: stream,
     options: {
       cwd,
+      model: process.env.GHOSTCLAW_MODEL || undefined,
       additionalDirectories: extraDirs.length > 0 ? extraDirs : undefined,
       resume: sessionId,
       resumeSessionAt: resumeAt,