Logging

Log Levels

Level	When to use	What you'll see
error	Something is broken and needs attention	Missing secrets, capability load failures, delivery failures
warn	Degraded but recovered	Fallback model activated, edit failed (sent as new message), retry succeeded
info	Normal operations	Boot phases, message delivery, tool calls, deploys
debug	Detailed flow	Prompt routing, session resolution, token counts
trace	Everything including swallowed errors	Every silent catch, failed probes, harmless timeouts

Set via LOG_LEVEL in .env. Default: info.

TRACE level

LOG_LEVEL=trace surfaces every error that's normally swallowed — including many harmless patterns:

• [platform_detect] spawnSync ETIMEDOUT — WSL journalctl probe timed out (sleep detection falls back to idle time)
• [phase_transport] pkill ... Command failed — no stale process to kill on clean boot (expected)
• [skill-watcher] ENOENT — optional skill file not present
• [irc-client] ECONNREFUSED — IRC server not running (non-critical service)

These are diagnostic noise at trace level, not bugs. Use TRACE when debugging a specific silent failure — don't leave it on permanently (log files grow fast).

Log location

~/.abtars/logs/bridge-YYYY-MM-DD.log

Rotated daily. Retention controlled by doctor.sh --fix (default: 7 days).

Useful grep patterns

# Errors only
grep "ERROR" ~/.abtars/logs/bridge-$(date +%F).log

# Swallowed errors (TRACE must be enabled)
grep "swallowed" ~/.abtars/logs/bridge-$(date +%F).log

# Message delivery
grep "Response delivered\|send failed" ~/.abtars/logs/bridge-$(date +%F).log

# Boot sequence
grep "✓\|✅\|ERROR" ~/.abtars/logs/bridge-$(date +%F).log | head -20