triggerdotdev/skills
6 skills10.7k installs162 starsGitHub
Install
npx skills add https://github.com/triggerdotdev/skillsSkills in this repo
1Trigger TasksTrigger Tasks is an agent skill covering advanced Trigger.dev v4 task patterns for indie SaaS backends. When you outgrow a single fire-and-forget job, you need tags for observability, batch triggers for bulk user events, and clarity on payload size and rate limits so production does not silently throttle. The skill walks through importing task and tags from @trigger.dev/sdk, adding execution-time tags, triggering with up to ten labels, and subscribing to runs by tag for progress UI or agent monitoring. Batch Triggering v2 sections spell out the 1,000-item ceiling, 3MB per-item payloads, and tier-specific refill rates—details agents often get wrong from memory. Use this while building notification pipelines, video processing queues, or multi-tenant workflows where org and user scoping must appear in run metadata. It assumes you already run Trigger.dev v4; it deepens task design rather than replacing baseline setup docs.2kinstalls2Trigger Configtrigger-config is an agent skill that teaches solo builders how to shape Trigger.dev’s trigger.config.ts end to end: project reference, task directories, runtime selection, logging, retries, global hooks, and build extensions that bundle ORMs and native deps for cloud execution. Indie SaaS teams often start with a hello-world task, then stall when Prisma migrations, connection strings, or retry semantics must be encoded in config rather than scattered env vars. This skill consolidates those choices into copy-ready TypeScript so your agent does not hallucinate deprecated keys. It is phase-specific to Build integrations—you use it while standing up or refactoring background work, paired with separate task-authoring skills for individual jobs. The readme emphasizes extension patterns starting with Prisma, which is the common path for Node SaaS stacks. After config is stable, you typically author tasks and deploy via Trigger.dev’s normal CLI flow; this skill does not replace infra monitoring in Operate.1.9kinstalls3Trigger Agentstrigger-agents teaches solo builders how to structure LLM work as reliable Trigger.dev tasks instead of one-off scripts. A decision table steers you toward parallelization when batching items, routing when models or handlers differ, prompt chaining when each step needs a quality gate, orchestrator-workers for specialized subtasks, evaluator-optimizer loops until output meets a bar, and human-in-the-loop via waitpoints when automation must pause for approval. Examples use the Trigger.dev SDK with the Vercel AI SDK generateText pattern so retries, durability, and observability stay in the platform layer. Use it when a feature needs tool calling against your own tasks, streamed progress to a frontend, or multi-step copy and translation pipelines. The skill is architectural: it names companion docs for streaming and ai.tool rather than replacing them, which fits indie teams shipping one agent feature at a time with room to grow into operate-phase monitoring of the same jobs.1.9kinstalls4Trigger SetupTrigger-setup is a focused environment guide for solo and indie builders wiring Trigger.dev into a Node/TypeScript project. It explains which secret key to copy from the dashboard, how development and production prefixes differ, and safe local patterns such as `.env` files excluded from version control or ephemeral shell exports before `npx trigger dev`. The skill extends into deployment hygiene: GitHub Actions and Vercel secret slots, plus staging and production key separation so preview deploys do not hit prod queues. Because tasks execute on Trigger.dev’s infrastructure, it also introduces syncing environment variables through config extensions rather than assuming your laptop’s `.env` is automatically present in the cloud. Use it when you are first connecting the SDK, rotating keys, or onboarding a new hosting environment—not when authoring task logic itself.1.8kinstalls5Trigger RealtimeTrigger-realtime is an agent skill for solo and indie builders who already run background work on Trigger.dev and need the product surface to feel alive instead of polling blindly. It documents how to mint scoped public access tokens on the server, pass read-only credentials to the browser, and optionally issue trigger public tokens so the client can start named tasks without leaking admin keys. On the backend it walks through tasks.trigger followed by runs.subscribeToRun so your agent or API layer can stream status transitions, metadata like progress percentages, and final output when a run reaches COMPLETED. The same patterns support React views that kick off jobs and watch them, long-running automation with human-in-the-loop approval steps, and streaming model responses through a task boundary. You reach for it during implementation when UX depends on sub-minute feedback, not after launch when you only need batch logs. It assumes TypeScript familiarity with @trigger.dev/sdk and a deployed Trigger.dev project; it does not replace defining tasks, retries, or deployment of workers themselves.1.8kinstalls6Trigger Cost Savingstrigger-cost-savings is a guidance skill for solo builders and small teams running background work on Trigger.dev. It compresses the vendor’s cost playbook into actionable patterns: monitor the usage dashboard for top compute hogs and retry drag, configure standard and spike billing alerts, start on small-1x machines and escalate presets only when CPU or memory proof demands it, and use trigger-time machine overrides for bursty jobs. Idempotency keys reduce duplicate billable executions when clients retry. The skill is procedural documentation rather than a code generator, and defers to the live docs URL for the freshest pricing details. Use while operating production task queues and when revisiting Ship-time task definitions that turned out oversized.1.3kinstalls