Pipeline Examples

Create a pipeline called "code-review-automation":

1. Trigger (manual) — accepts code diff as payload
2. Agent (reviewer) — "Review this code for bugs, security issues, and style violations: {{input}}"
3. Condition — check if {{input}} contains "issue" or "warning"
4. True branch → Skill (slack-notify) — post review results to #code-reviews
5. False branch → Output (log) — "No issues found"

Create a pipeline called "daily-standup":

1. Trigger (schedule, cron: "0 9 * * 1-5") — weekdays at 9am
2. Agent (manager) — "Generate a daily standup report: what was completed yesterday, what's in progress today, and any blockers"
3. Transform — format the report as a Slack message
4. Skill (slack-notify) — post to #standup channel
5. Memory Store — save the standup report for future reference

Create a pipeline called "api-health-check":

1. Trigger (schedule, cron: "*/15 * * * *") — every 15 minutes
2. HTTP Request — GET https://api.yourapp.com/health
3. Code — parse response and check status:
   const data = JSON.parse(input);
   return { healthy: data.status === "ok", responseTime: data.responseTime };
4. Condition — {{input.healthy}} equals true
5. True branch → Output (log) — "API healthy"
6. False branch → Skill (telegram-notify) — "API is down! Response: {{input}}"

Create a pipeline called "github-pr-review":

1. Trigger (webhook, path: "/hooks/github-pr", method: POST)
2. Code — extract PR details:
   const payload = JSON.parse(input);
   return { title: payload.pull_request.title, diff: payload.pull_request.body, url: payload.pull_request.html_url };
3. Agent (analyst) — "Analyze the impact of this PR: {{input.title}}. Changes: {{input.diff}}"
4. Agent (reviewer) — "Review this PR for code quality and security: {{input.diff}}"
5. Merge — combine analyst and reviewer outputs
6. Transform — format as a GitHub comment
7. HTTP Request — POST comment back to GitHub PR via API

Create a pipeline called "knowledge-sync":

1. Trigger (file_change, glob: "docs/**/*.md", debounceMs: 3000)
2. Code — read and chunk the changed file content
3. Loop — iterate over chunks
4. Memory Store — store each chunk with metadata: {"source": "docs", "file": "{{vars.filename}}"}
5. Aggregate — collect stored memory IDs
6. Output (log) — "Synced {{input.length}} memory chunks from docs"

Create a pipeline called "incident-response":

1. Trigger (webhook, path: "/hooks/alert")
2. Agent (analyst) — "Analyze this alert and determine severity: {{input}}"
3. Switch — route by severity:
   - "critical" → parallel: Skill (telegram-notify) + Agent (developer, "Generate a hotfix plan")
   - "warning" → Skill (slack-notify) with warning message
   - default → Output (log)
4. Memory Store — store incident details for post-mortem

Create a pipeline called "content-pipeline":

1. Trigger (manual) — topic as payload
2. Agent (researcher) — "Research the topic: {{input}}. Find key points, statistics, and examples"
3. Agent (developer) — "Write a blog post based on this research: {{input}}"
4. Agent (reviewer) — "Review this article for accuracy, readability, and SEO: {{input}}"
5. Condition — check if review passes quality threshold
6. True branch → Output (file, path: "content/{{vars.slug}}.md")
7. False branch → Agent (developer) — "Revise the article based on this feedback: {{input}}"