examples

# Examples & Patterns Ready-to-use examples and complete workflow scenarios. ## MCP Tool Examples | File | Description | |------|-------------| | @references/mcp/tool-patterns.md | Common MCP tool call patterns and workflows (tasks, runs, TRQL, prompts, health report, agent chat, session channels) | ## Scenario Walkthroughs | File | Description | |------|-------------| | @references/scenarios/webhook-processor.md | Complete webhook processing task with queue and retry | | @references/scenarios/ai-pipeline.md | AI agent pipeline with chaining, tools, and streaming | | @references/scenarios/cron-data-sync.md | Scheduled data sync with batch processing | ## Quick Reference Patterns ### Initialize New Project ``` 1. list_orgs() → pick org 2. initialize_project(orgParam, projectName, cwd) → setup guide (writes no files); follow it 3. get_current_worker(environment="dev") → verify ``` ### Trigger and Monitor ``` 1. get_current_worker() → find the task slug 2. get_task_schema(taskSlug) → payload schema 3. trigger_task(taskId, payload) → get run ID 4. wait_for_run_to_complete(runId) → get result ``` ### Debug Failures ``` 1. list_runs(status="FAILED", period="1d") 2. get_run_details(runId) → read trace + span IDs 3. get_span_details(runId, spanId) → AI enrichment (model, tokens, cost) if GenAI span 4. Fix code, redeploy 5. trigger_task(taskId, payload) → retry ``` ### Deploy to Production ``` 1. deploy(environment="staging") → test first 2. trigger_task(environment="staging") → verify 3. deploy(environment="prod") → ship it 4. list_deploys(environment="prod", limit=1) → confirm ``` ### Run Dev Server from an Agent ``` 1. start_dev_server(configPath?) → launches `trigger dev` in background 2. dev_server_status(lines=50) → poll until status="ready" 3. get_current_worker(environment="dev") → task list 4. trigger_task(...) → test task 5. stop_dev_server() → tear down ``` ### Analyze Failures with TRQL ``` 1. get_query_schema(table="runs") 2. query({ query: "SELECT task_identifier, count() AS n FROM runs WHERE status = 'Failed' GROUP BY task_identifier ORDER BY n DESC LIMIT 10", period: "7d" }) ``` ### LLM Cost Breakdown ``` 1. get_query_schema(table="llm_metrics") 2. query({ query: "SELECT request_model AS model, sum(total_cost) AS usd FROM llm_metrics GROUP BY model ORDER BY usd DESC", period: "30d" }) ``` ### Health Check ``` get_report(key="health", environment="prod", period="24h") → verdict + next action ``` ### Switch Profile Mid-Session ``` 1. whoami() → current profile + API URL 2. list_profiles() → available profiles 3. switch_profile("self-hosted-prod") → subsequent tool calls use new profile ``` ### Prompt Override Hotfix ``` 1. list_prompts(environment="prod") → find slug 2. get_prompt_versions(slug, environment="prod") → see current/override state 3. create_prompt_override(slug, textContent, commitMessage) → apply hotfix 4. (iterate) update_prompt_override(slug, textContent) 5. remove_prompt_override(slug) → revert after code fix deploys ```