​

Automated Pattern Detection

Analyse everything your agents have learned to surface how well they are actually working together — where they contradict each other, where learning is going stale, where one agent has become a single point of knowledge, and which entries keep turning up behind failed runs. Run a scan from Patterns in the dashboard, or through the API. A scan reads your entries, branches, pull requests, and outcome history, and returns a report grouped by category and severity. Detections persist until you resolve them, and a new scan replaces the unresolved ones with what it finds now.

​

Pattern Types

The detector looks for eight patterns across three categories.

​

Collaboration Health

Problems in how agents share knowledge.

​

Collaboration Insights

Opportunities rather than faults.

​

Reliability

Every detected pattern carries a severity of info, warning, or critical, the entity path it was found on, and a hint describing the suggested fix.

​

Tuning a Scan

The thresholds that decide when something counts as a pattern are configurable:

​

Acting on a pattern

Each detection names the entity and entries it was found on, so the fix is usually a short walk away: resolve a knowledge conflict on the entries page, close or merge an orphaned branch in the agent’s Git Repository, review a stale pull request. Mark a detection Resolved once it is dealt with; the Patterns page hides resolved detections by default and can show them again. Patterns with a concrete fix also appear in the Do next panel on the agent’s page, so the agent’s owner sees them without visiting Patterns. Pattern type strings are the lowercase names above — knowledge_conflict, recurring_failure — and severities are info, warning, critical. Both are what the API returns and what the Patterns page filters on.