Pipeline Triggers

Triggers define when a pipeline runs. You can combine multiple trigger types or run pipelines manually.

Manual Trigger

The simplest trigger — run a pipeline on demand via the desktop app or MCP tools.

pipeline_run id="my-pipeline"

No configuration needed. Every pipeline supports manual execution regardless of other triggers.

Schedule (Cron)

Run pipelines on a recurring schedule using standard cron expressions.

{
  "type": "schedule",
  "cron": "0 9 * * 1-5",
  "timezone": "America/New_York"
}

Common Cron Patterns

Use Cases

• Daily standup report — Aggregate tasks, generate summary, post to Slack
• Hourly health check — Hit your API, alert on failures
• Weekly code review — Scan for stale PRs, notify team

Webhook

Expose an HTTP endpoint that starts a pipeline when called. The request body is passed as the trigger payload.

{
  "type": "webhook",
  "path": "/hooks/deploy",
  "method": "POST",
  "secret": "your-webhook-secret"
}

Use Cases

• GitHub PR webhook — Trigger code review pipeline on new PRs
• Deployment webhook — Run post-deploy checks after CI/CD
• Form submission — Process incoming data from external services

Example: GitHub PR Review

1. Create a pipeline with a webhook trigger at /hooks/github-pr
2. Add GitHub webhook pointing to http://your-server:3000/hooks/github-pr
3. Pipeline receives PR data → Agent reviews → Posts results back

Event

Listen to internal EventBus events. Triggers when a matching event is emitted within Agents Machine.

{
  "type": "event",
  "eventName": "memory:stored",
  "filter": "category == 'architecture'"
}

Available Events

Use Cases

• Auto-tag memories — When a memory is stored, run classification agent
• Task notifications — When a task moves to "done", notify the team
• Audit logging — When a secret is accessed, log the event

File Change

Watch filesystem paths and trigger when files are created, modified, or deleted.

{
  "type": "file_change",
  "glob": "src/**/*.ts",
  "debounceMs": 2000
}

Use Cases

• Auto-review on save — Watch source files, run code review agent on changes
• Documentation sync — Watch markdown files, update memory when docs change
• Test runner — Watch test files, run tests on modification

Combining Triggers

A pipeline can have multiple triggers. For example, a code review pipeline might be:

• Webhook — Triggered by GitHub on new PRs
• Schedule — Daily scan of all open PRs at 9am
• Manual — On-demand review when needed

The trigger type is recorded in each pipeline run for audit purposes.