Watchdog
Configure monitoring rules that watch your logs and fire alerts or kick off automatic analyses
Watchdog rules continuously evaluate your project's logs against thresholds you configure. When a rule trips, you either get a notification or a full automatic analysis — your choice, per rule.
Use these endpoints to manage rules programmatically: list and create
them, dry-run a rule against the last seven days of logs to estimate
alert volume, and pull the recent activity feed. Toggle individual
rules with PUT /v1/config/monitoring/{id}.
The project-wide master switch lives in the dashboard at Project →
Watchdog → Settings.
Endpoints
| Method + path | Purpose |
|---|---|
GET /v1/config/monitoring | List rules for the project |
POST /v1/config/monitoring | Create a rule |
PUT /v1/config/monitoring/{id} | Update a rule |
DELETE /v1/config/monitoring/{id} | Delete a rule |
POST /v1/config/monitoring/dry-run | Replay a rule against historical logs |
GET /v1/config/monitoring/activity | Recent rule activity and alerts |
Rule shape
Every rule is one of four types — error count, error rate, pattern match, or spike — paired with a check schedule, a lookback window, and an action (notify or auto-analyze). The full field-by-field breakdown lives in the create endpoint and a conceptual walkthrough lives in Rule types.
See also
- Conceptual walkthrough — Watchdog Deep-dive
- Rule types — error-count, error-rate, pattern, and spike detection explained
- Scheduling — fixed vs. randomized intervals and how to pick one
- Alert behavior — cooldowns, daily caps, deduplication
- Cost calculator — estimating analysis usage from rule cadence
- Dry-run — replaying a rule against past logs before enabling it
- Activity feed — what gets recorded for every check
- Authentication — required permissions per endpoint