Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.wednesdayai.dev/llms.txt

Use this file to discover all available pages before exploring further.

​
Logging

WednesdayAI logs in two places:
  • File logs — JSON lines written by the gateway to a rolling log file.
  • Console output — structured, colorized output in terminals and the control panel.

​
Where logs live

The gateway writes a rolling log file at: 
/tmp/openclaw/openclaw-YYYY-MM-DD.log
/tmp is periodically purged by the OS. For persistent log retention, override the log path to a stable location.
Override the log path in ~/.openclaw/openclaw.json: 
{
  logging: {
    file: "~/.openclaw/logs/openclaw.log",
  },
}

​
Reading logs

openclaw logs --follow
The CLI tails the gateway log file via RPC and formats entries in real time. Output flags:
FlagEffect
--jsonLine-delimited JSON, one event per line
--plainForce plain text (no colors)
--no-colorDisable ANSI colors in TTY mode
In --json mode each line has a type field:
  • meta — stream metadata (file path, cursor, size)
  • log — parsed log entry
  • notice — truncation or rotation hints
  • raw — unparsed line (malformed JSON)

​
Control panel

The control panel’s Logs tab tails the same file. Access it at http://localhost:18789/ → Logs.

​
Channel-specific logs

openclaw channels logs --channel whatsapp
openclaw channels logs --channel telegram --follow

​
Systemd journal (Linux)

If the gateway runs as a systemd service: 
journalctl --user -u openclaw-gateway -f
journalctl --user -u openclaw-gateway --since "1 hour ago"

​
Configuration

All logging settings live under logging in ~/.openclaw/openclaw.json. Most changes apply immediately (hot-reload); log file path changes require a restart. 
{
  logging: {
    level: "info",              // file log verbosity
    file: "/tmp/openclaw/openclaw-YYYY-MM-DD.log",
    consoleLevel: "info",       // console verbosity
    consoleStyle: "pretty",     // pretty | compact | json
    redactSensitive: "tools",   // off | tools
    redactPatterns: ["sk-.*"],  // additional regex patterns to redact
  },
}

​
Log levels

Available levels (least to most verbose): error → warn → info → debug → trace Precedence (first wins):
  1. --log-level <level> CLI flag
  2. OPENCLAW_LOG_LEVEL env var
  3. logging.level / logging.consoleLevel in config
To temporarily raise verbosity for a single command without editing config: 
OPENCLAW_LOG_LEVEL=debug openclaw gateway status
openclaw --log-level debug gateway run
--verbose affects only console output; it does not change file log levels.

​
Console styles

StyleDescription
prettyColorized, timestamped, human-readable (default)
compactTighter output — good for long sessions
jsonJSON per line — for log processors and pipelines

​
Redaction

redactSensitive: "tools" (default) redacts sensitive-looking values from tool call summaries in console output. File logs and OTLP exports are not affected. Add custom patterns to redactPatterns to redact additional secrets: 
{
  logging: {
    redactPatterns: ["sk-.*", "Bearer .*", "token=[A-Za-z0-9]+"],
  },
}

​
OpenTelemetry export

WednesdayAI can export telemetry — metrics, traces, and logs — via OTLP/HTTP to any OpenTelemetry-compatible collector (Grafana, Jaeger, Honeycomb, Datadog, etc.).

​
Enable diagnostics

{
  diagnostics: {
    enabled: true,
    otel: {
      endpoint: "http://otel-collector:4318",  // your OTLP/HTTP endpoint
      sampleRate: 1.0,                          // 0.0–1.0, root spans only
      flushIntervalMs: 5000,                    // min 1000ms
      logs: false,                              // true to export logs over OTLP
    },
  },
}
Or via standard OpenTelemetry env vars: 
OTEL_EXPORTER_OTLP_ENDPOINT=http://otel-collector:4318

​
What gets exported

Metrics (counters + histograms):
MetricDescription
openclaw.message.queuedMessages enqueued per channel
openclaw.message.processedMessages processed, by outcome
openclaw.message.duration_msProcessing latency histogram
openclaw.model.usageToken usage, cost, latency per model/provider
openclaw.webhook.receivedWebhook ingress per channel
openclaw.queue.lane.*Queue depth, wait time, task completion
openclaw.session.stuckSessions stuck in a state
Traces: openclaw.model.usage, openclaw.webhook.processed, openclaw.message.processed Logs: exported over OTLP when diagnostics.otel.logs: true. Console redaction does not apply to OTLP log export.

​
Sampling and flushing

  • diagnostics.otel.sampleRate: 0.0 (none) to 1.0 (all). Applies to root spans only.
  • diagnostics.otel.flushIntervalMs: how often metrics are flushed. Minimum 1000ms.
  • High-volume installs should use a local OTel collector with sampling/filtering.

​
Troubleshooting

  1. Confirm the gateway is running: openclaw gateway status
  2. Check the configured file path: openclaw doctor
  3. Check write permissions on the log directory
  4. Verify the gateway process has started since the last config change
The CLI cannot connect to the gateway RPC socket.
openclaw doctor     # diagnose connectivity and config issues
openclaw gateway status
If the gateway is running but logs still fail, check that gateway.rpc is not disabled in config.
Temporarily raise the log level without editing config:
OPENCLAW_LOG_LEVEL=debug openclaw gateway --port 18789
Or for a running daemon:
# Edit config hot-reload
jq '.logging.level = "debug"' ~/.openclaw/openclaw.json > /tmp/oc.json && mv /tmp/oc.json ~/.openclaw/openclaw.json
The gateway watches the config file and applies level changes immediately. No restart needed.
  1. Confirm diagnostics are enabled: openclaw doctor
  2. Check the endpoint is reachable from the gateway host: curl -v http://otel-collector:4318/v1/metrics
  3. Check for OTLP errors in the gateway log: openclaw logs --follow | grep otlp
  4. If the endpoint already includes /v1/traces or /v1/metrics, it is used as-is. Don’t double-append the path.