Published 2026-09-21 · Tested 2026-08-19

Trigger.dev

A

Trigger.dev received 11 PASS votes and passed 5 of five agent surface checks. The clearest finding came from the find the exact limits task.

Panel: GPT 5.6 Sol, Opus 5, DeepSeek v4F Battery: v1 Read as markdown (opens in a new tab)

Three AI models, GPT 5.6 Sol, Claude Opus 5, and DeepSeek v4 Flash, each read Trigger.dev’s public documentation independently and attempted five first-hour developer jobs: run a background task, find the exact limits, recover from throttling, verify an alert webhook, use the TypeScript SDK.

No accounts, API calls, or code execution were used. Every verdict came from public pages and every published quotation passed a live verification check.

Freshness

How rechecks work
Tested
Quotes verified
Surface rechecked

No change since the test.

5 of 5 quoted passages still appear on the live pages.

Trigger.dev Mintlify · published
A

95.0% · 76/80 · AI Agent Readiness Score · reading 30 pts · surface 50 pts

llms.txt PASS
llms-full.txt PASS
markdown mirror PASS
MCP server PASS
docs AI PASS
Task GPT 5.6 SolOpus 5DeepSeek v4F Consensus
Run a background task PASSPASSPASS PASS
Find the exact limits PARTIALPARTIALPARTIAL PARTIAL
Recover from throttling PASSPASSPASS PASS
Verify an alert webhook PASSPASSPASS PASS
Use the TypeScript SDK PARTIALPASSPASS PASS

docs platform: Mintlify (unscored) · verified 2026-08-19

What to fix first

This fix could add up to 3 points to the AI Agent Readiness Score. The list ranks each fix by the points it would add. How the ranking works

  1. 1
    +3 points Find the exact limits PARTIAL

    Found: The Limits page gives a 3MB payload and 10MB output cap, while the Triggering page says 10MB and 100MB.

    Fix: Use one payload cap and one output cap on both the Limits and Triggering pages.

    Evidence: trigger.dev/docs/triggering (opens in a new tab)

What the docs get right

  • Run a background task: 3 PASS votes. The quick start covers account creation, CLI initialization, Hello World, development mode, dashboard testing, and run confirmation.
  • Recover from throttling: 3 PASS votes. The SDK exposes isRateLimited and retryAfterMs, and its bounded example waits before retrying or rethrows other errors.
  • 5 of 5 agent surface checks. Present: llms.txt, llms-full.txt, markdown mirrors, an MCP server, docs AI.

Run a background task

PASS

PASS consensus from 3 PASS.

The quick start covers account creation, CLI initialization, Hello World, development mode, dashboard testing, and run confirmation. One page carries the whole path, npx trigger.dev@latest init, the Hello World example task, npx trigger.dev@latest dev, and the dashboard Test button, with the manual route split off to a separately labeled Manual setup page. The quick-start page names every command in order, init, dev, dashboard Test page, and the init command scaffolds the Hello World example task automatically.

Find the exact limits

PARTIAL

PARTIAL consensus from 3 PARTIAL.

The limits page covers every category, but Triggering conflicts on payload and output caps. The Limits page gives per-plan numbers for concurrency, queue depth, API rate, batch buckets, retention, schedules, and alerts in one table set, but /docs/triggering states a 10MB payload limit and a 100MB output limit against the Limits page's 3MB and 10MB. The Limits page gives per-plan numbers for concurrency, queue depth, API rate, batch buckets, retention, schedules, and alerts in one table set, but the pricing FAQ contradicts the Pro concurrency limit (100+ vs 200) and the triggering page gives a different single-payload cap (3MB vs 10MB).

Recover from throttling

PASS

PASS consensus from 3 PASS.

The SDK exposes isRateLimited and retryAfterMs, and its bounded example waits before retrying or rethrows other errors. The condition is a typed error with isRateLimited and retryAfterMs, the correction is a working retry loop that sleeps for retryAfterMs, and the token-bucket sizes and refill rates are tabled per plan on the Limits page.

Verify an alert webhook

PASS

PASS consensus from 3 PASS.

The alert guide covers creation, run-failure selection, alert.run.failed, secret lookup, verification, and invalid-signature handling. Creation, the alert.run.failed switch case, webhooks.constructEvent with the secret, WebhookError handling, and the full run-failure payload schema are all on one page, though constructEvent appears nowhere else in the docs and has no reference entry. The alerts page documents creation, the alert.run.failed event type, the full payload schema, and tells you to use the SDK to parse the webhook. The SDK's webhooks.constructEvent() pattern is the natural completion, though constructEvent has no dedicated reference page.

Use the TypeScript SDK

PASS

PASS consensus from 2 PASS, 1 PARTIAL.

The SDK form is clear, but Tasks Overview imports helloWorld after defining it without export. @trigger.dev/sdk is named consistently across quick start, manual setup, tasks overview, and the management API, and the task({ id, run }) form is identical in every example. @trigger.dev/sdk is the single package name across quick start, manual setup, tasks overview, and triggering pages, and the task({ id, run }) form is identical in every code example.

The receipt

We currently have a hard limit on task payloads above 10MB.

The limits page covers every category, but Triggering conflicts on payload and output caps.

Agent surface notes

Initialize returned JSON-RPC protocol 2025-06-18 and server name Trigger.dev.

The live docs expose assistant-entry controls labeled Toggle assistant panel.

Show the score

AI Agent Readiness Score 95.0%, grade A

Paste this into a readme:

[![AI Agent Readiness Score 95.0%](https://docsforagents.com/badge/trigger-dev.svg)](https://docsforagents.com/reports/trigger-dev-docs-ai-agent-readiness/)

Method note

This is a reading test of public documentation, not an execution test. No accounts were created and no API calls were run. The AI Agent Readiness Score counts fifteen reading votes at PASS 2, PARTIAL 1, and FAIL 0, for 30 possible points. Five agent surface checks add 10 points each. The total is 80. Consensus chips show each row majority and do not affect scoring. The panel split on 1 of five tasks. Quotes shown here were re-fetched and confirmed verbatim on 2026-08-19.

Read the full methodology

Put another docs site through the battery.

Nominate a docs site