nthfuncx/nanoclaw
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

docs: update skills to use Docker commands after runtime migration (#325)

Browse Source 
All skills now reference Docker CLI instead of Apple Container CLI.
Setup skill defaults to Docker with optional /convert-to-apple-container.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
gavrielc
2026-02-20 14:06:31 +02:00
committed by GitHub
parent a7faac664a
commit 6b9b3a12c9
7 changed files with 41 additions and 161 deletions
Download Patch File Download Diff File Expand all files Collapse all files

107
.claude/skills/add-voice-transcription/SKILL.md
View File

@@ -296,86 +296,20 @@ if (registeredGroups[chatJid]) {
}
```
### Step 6: Fix Orphan Container Cleanup (CRITICAL)
**This step is essential.** When the NanoClaw service restarts (e.g., `launchctl kickstart -k`), the running container is detached but NOT killed. The new service instance spawns a fresh container, but the orphan keeps running and shares the same IPC mount directory. Both containers race to read IPC input files, causing the new container to randomly miss messages — making it appear like the agent doesn't respond.
The existing cleanup code in `ensureContainerSystemRunning()` in `src/index.ts` uses `container ls --format {{.Names}}` which **silently fails** on Apple Container (only `json` and `table` are valid format options). The catch block swallows the error, so orphans are never cleaned up.
Find the orphan cleanup block in `ensureContainerSystemRunning()` (the section starting with `// Kill and clean up orphaned NanoClaw containers from previous runs`) and replace it with:
```typescript
// Kill and clean up orphaned NanoClaw containers from previous runs
try {
const listJson = execSync('container ls -a --format json', {
stdio: ['pipe', 'pipe', 'pipe'],
encoding: 'utf-8',
});
const containers = JSON.parse(listJson) as Array<{ configuration: { id: string }; status: string }>;
const nanoclawContainers = containers.filter(
(c) => c.configuration.id.startsWith('nanoclaw-'),
);
const running = nanoclawContainers
.filter((c) => c.status === 'running')
.map((c) => c.configuration.id);
if (running.length > 0) {
execSync(`container stop ${running.join(' ')}`, { stdio: 'pipe' });
logger.info({ count: running.length }, 'Stopped orphaned containers');
}
const allNames = nanoclawContainers.map((c) => c.configuration.id);
if (allNames.length > 0) {
execSync(`container rm ${allNames.join(' ')}`, { stdio: 'pipe' });
logger.info({ count: allNames.length }, 'Cleaned up stopped containers');
}
} catch {
// No containers or cleanup not supported
}
```
### Step 7: Build and Restart
### Step 6: Build and Restart
```bash
npm run build
```
Before restarting the service, kill any orphaned containers manually to ensure a clean slate:
```bash
container ls -a --format json | python3 -c "
import sys, json
data = json.load(sys.stdin)
nc = [c['configuration']['id'] for c in data if c['configuration']['id'].startswith('nanoclaw-')]
if nc: print(' '.join(nc))
" | xargs -r container stop 2>/dev/null
container ls -a --format json | python3 -c "
import sys, json
data = json.load(sys.stdin)
nc = [c['configuration']['id'] for c in data if c['configuration']['id'].startswith('nanoclaw-')]
if nc: print(' '.join(nc))
" | xargs -r container rm 2>/dev/null
echo "Orphaned containers cleaned"
```
Now restart the service:
```bash
launchctl kickstart -k gui/$(id -u)/com.nanoclaw
```
Verify it started with exactly one (or zero, before first message) nanoclaw container:
Verify it started:
```bash
sleep 3 && launchctl list | grep nanoclaw
container ls -a --format json | python3 -c "
import sys, json
data = json.load(sys.stdin)
nc = [c for c in data if c['configuration']['id'].startswith('nanoclaw-')]
print(f'{len(nc)} nanoclaw container(s)')
for c in nc: print(f' {c[\"configuration\"][\"id\"]} - {c[\"status\"]}')
"
```
### Step 8: Test Voice Transcription
### Step 7: Test Voice Transcription
Tell the user:
@@ -431,41 +365,6 @@ The architecture supports multiple providers. To add Groq, Deepgram, or local Wh
## Troubleshooting
### Agent doesn't respond to voice messages (or any messages after a voice note)
**Most likely cause: orphaned containers.** When the service restarts, the previous container keeps running and races to consume IPC messages. Check:
```bash
container ls -a --format json | python3 -c "
import sys, json
data = json.load(sys.stdin)
nc = [c for c in data if c['configuration']['id'].startswith('nanoclaw-')]
print(f'{len(nc)} nanoclaw container(s):')
for c in nc: print(f' {c[\"configuration\"][\"id\"]} - {c[\"status\"]}')
"
```
If you see more than one running container, kill the orphans:
```bash
# Stop all nanoclaw containers, then restart the service
container ls -a --format json | python3 -c "
import sys, json
data = json.load(sys.stdin)
running = [c['configuration']['id'] for c in data if c['configuration']['id'].startswith('nanoclaw-') and c['status'] == 'running']
if running: print(' '.join(running))
" | xargs -r container stop 2>/dev/null
container ls -a --format json | python3 -c "
import sys, json
data = json.load(sys.stdin)
nc = [c['configuration']['id'] for c in data if c['configuration']['id'].startswith('nanoclaw-')]
if nc: print(' '.join(nc))
" | xargs -r container rm 2>/dev/null
launchctl kickstart -k gui/$(id -u)/com.nanoclaw
```
**Root cause:** The `ensureContainerSystemRunning()` function previously used `container ls --format {{.Names}}` which silently fails on Apple Container (only `json` and `table` formats are supported). Step 6 of this skill fixes this. If you haven't applied Step 6, the orphan problem will recur on every restart.
### "Transcription unavailable" or "Transcription failed"
Check logs for specific errors: