Skip to content

Trigger.dev Skills - Setup Guide

Source: triggerdotdev/skills via skills.sh - 14.6K combined installs across 13 indexed listings; first seen Oct 9, 2026 (extended-query probe) GitHub: triggerdotdev/skills (33 stars, no LICENSE file; pushed 2026-09-01; branch main; skill directories at the repo root, e.g. trigger-tasks/SKILL.md) Category: Background Jobs / Realtime Infrastructure Quality Tier: 🟡 Beta - official Trigger.dev org; no LICENSE file in repo; small stars (33); all sampled verdicts Pass

Trigger.dev Skills is the official skills collection from Trigger.dev, a durable execution platform for AI agents, workflows, and background tasks. The skills cover the full arc of a Trigger.dev project: bootstrapping, backend task authoring, frontend realtime consumption, configuration and build extensions, AI agent patterns, and cost optimization.

The repo is an automatic mirror: the skills are maintained in the Trigger.dev monorepo and ship inside the @trigger.dev/sdk package, and a sync workflow republishes them here so they install through skills.sh. Start with trigger-setup to bootstrap a project, then move to trigger-tasks once the /trigger directory exists; trigger-realtime covers the consumer side for React and Next.js frontends.

Installation

skills.sh

# Install all skills
npx skills add triggerdotdev/skills

# Install a single skill
npx skills add triggerdotdev/skills --skill trigger-setup

Already on Trigger.dev?

The same skills ship inside the SDK and can be installed straight into your coding agent:

npx trigger.dev@latest install-mcp

The repo is an automatic mirror - contribute skill changes in the Trigger.dev monorepo and they flow here automatically.

What It Provides

Skill Installs Use For
trigger-tasks 2,827 Writing backend tasks with @trigger.dev/sdk: task() and schemaTask(), retries, waits, queues, concurrency, idempotency, and scheduled tasks
trigger-setup 2,504 Bootstrapping Trigger.dev into an existing project: CLI auth, SDK install, trigger.config.ts, a first task, and the dev server
trigger-realtime 2,473 Frontend realtime: subscribing to runs, React hooks, browser triggers, and scoped public tokens
trigger-config 2,253 trigger.config.ts configuration: build extensions for Prisma, Playwright, FFmpeg, and Python, plus deployment settings
trigger-agents 2,218 AI agent patterns: orchestration, parallelization, routing, evaluator-optimizer, and human-in-the-loop workflows
trigger-cost-savings 1,957 Cost optimization: static source analysis plus live run analysis via the Trigger.dev MCP tools

The remaining 7 indexed listings range from 3 to 185 installs (trigger-authoring-chat-agent 185; trigger-chat-agent-advanced 181; tasks, setup, realtime, config, and agents at 3 each).

Why This Matters for Hermes Agents

Background jobs and durable workflows are exactly the workloads agents get asked to build, and Trigger.dev's guarantees - retries with backoff, queues, long waits, realtime updates - are easy to use incorrectly without current SDK guidance. These skills put the platform's own best practices in front of the agent at the moment it writes task code, from task() and schemaTask() definitions to idempotency keys and scheduled tasks. The realtime skill covers the consumer half that agents routinely botch: subscribing to runs, streaming into React, and minting scoped public tokens instead of leaking secret keys. The agent-pattern skill packages orchestration, routing, evaluator-optimizer, and human-in-the-loop designs on top of durable execution, a cleaner substrate for multi-step agents than ad-hoc loops. The cost skill even audits spend with live run data through Trigger.dev's MCP tools. And because the skills ship inside the SDK, they track the version you actually deploy; the mirror here is just a distribution channel.

Usage

You say What happens
Add Trigger.dev to my existing Next.js app trigger-setup authenticates the CLI, installs the SDK, writes trigger.config.ts, and scaffolds /trigger with a first task
Write a durable task that retries with backoff trigger-tasks covers task() and schemaTask(), retries, waits, queues, and idempotency keys
Show live run progress in my React UI trigger-realtime wires useRealtimeRun and useRealtimeStream with scoped public tokens
Configure build extensions for Prisma and Playwright trigger-config covers trigger.config.ts, build extensions, and deployment settings
Build a multi-step agent with approval gates trigger-agents applies orchestration and human-in-the-loop patterns on durable execution
Reduce our Trigger.dev spend trigger-cost-savings audits tasks, schedules, and runs with static analysis plus live run data

Verification

Confirm the install from the CLI or the skills directory:

npx skills list | grep trigger

The skills install as trigger-* directories at the repo root (trigger-tasks/SKILL.md layout). Before installing, review the exact instructions the agent will read: fetch the tasks skill file at https://raw.githubusercontent.com/triggerdotdev/skills/main/trigger-tasks/SKILL.md (verified reachable Oct 10, 2026).

Security

skills.sh verdicts for sampled skills (verified Oct 10, 2026) - re-check the security pages on skills.sh before production use:

Skill Gen Agent Trust Hub Socket Snyk
trigger-tasks Pass Pass Pass
trigger-setup Pass Pass Pass
trigger-realtime Pass Pass Pass

Integration with CorpusIQ

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects

Limitations

  • No LICENSE file in the repository; the skills ship inside the @trigger.dev/sdk package, but redistribution terms are not stated here - confirm with Trigger.dev before vendoring.
  • The repo is an automatic mirror - edit changes in the Trigger.dev monorepo; content can change here without notice as the sync runs.
  • The install index outruns the tree: trigger-config and trigger-agents carry install counts but have no matching directories in the current mirror, and five generic listings (tasks, setup, realtime, config, agents) sit at 3 installs each.
  • Verdicts cover the three sampled skills only; the chat-agent skills and the cost skill are unaudited here.

  • The small star count (33) reflects the mirror role rather than adoption - installs arrive through the SDK and skills.sh, not GitHub stars.

  • Snapshot data, verified Oct 10, 2026: 14,613 combined installs across 13 indexed listings; 33 GitHub stars; no LICENSE file; last pushed 2026-09-01. Counts drift over time.

Pro Tips

  • Bootstrap first: run trigger-setup and get a dev server running before asking for task code, then switch to trigger-tasks.
  • Verify against the live platform: tasks are easiest to test with the dev server, and the realtime hooks need a running run to observe.
  • Keep secrets server-side; use auth.createPublicToken for frontend subscriptions instead of exposing the secret key.
  • Ask for cost reviews periodically; trigger-cost-savings combines static analysis with live data from the Trigger.dev MCP tools.
  • Remember the mirror rule: skill fixes belong in the Trigger.dev monorepo, and they flow back here automatically.

← Skills Catalog | Skills Marketplace → Powered by CorpusIQ

AI answers on your data. See where every number comes from.

Connect your tools and ask in ChatGPT, Claude, or Perplexity. Enterprise-grade answers in minutes. From $29.95/month.

Start free, no credit card
Contact Terms Privacy Pricing