Product MCP Server

TrigRun's MCP server lets AI agents manage scheduled jobs, inspect executions, configure notifications, and handle secrets — all through natural language or direct tool calls.

Quick start

Step 1 — Get an API token

Log in to trigrun.com, open API Keys, and create a token. Copy the value right away because the full token is only shown once. It starts with cron_pat_, and you'll pass it as the token argument in every tool call.

If you need a step-by-step walkthrough, see the Authentication guide.

Step 2 — Add the MCP server

TrigRun exposes a hosted Streamable HTTP MCP endpoint:

https://api.trigrun.com/mcp

Point any Streamable HTTP MCP client at that URL.

Cursor or any remote Streamable HTTP client:

{
  "mcpServers": {
    "trigrun": {
      "url": "https://api.trigrun.com/mcp"
    }
  }
}

If your client asks for a transport, choose Streamable HTTP.

Step 3 — Verify it works

Ask your agent:

List my TrigRun jobs using token cron_pat_your_token_here.

The agent will call list_jobs and show your existing jobs (or an empty list if you're new).

Authentication

Every tool requires a token argument. Use an API token for long-lived agent access:

{
  "tool": "list_jobs",
  "arguments": {
    "token": "cron_pat_abc123..."
  }
}

You can also pass a short-lived JWT from POST /v1/auth/login if you prefer session-based auth.

Available tools

Jobs

list_jobs

List all scheduled jobs in your workspace. Optionally filter by state.

{
  "tool": "list_jobs",
  "arguments": {
    "token": "cron_pat_abc123...",
    "state": "active"
  }
}

get_job

Get full details of a specific job by ID.

{
  "tool": "get_job",
  "arguments": {
    "token": "cron_pat_abc123...",
    "job_id": "clx9abc123def"
  }
}

create_job

Create a new scheduled HTTP job. Supports the schedule kinds exposed by the CLI and dashboard.

TrigRun evaluates cron expressions in UTC. Agents should convert local wall-clock schedules to UTC before creating jobs.

{
  "tool": "create_job",
  "arguments": {
    "token": "cron_pat_abc123...",
    "name": "hourly-sync",
    "url": "https://api.example.com/sync",
    "method": "POST",
    "kind": "cron",
    "cron": "0 * * * *",
    "headers": { "X-Source": "trigrun" }
  }
}

{
  "tool": "create_job",
  "arguments": {
    "token": "cron_pat_abc123...",
    "name": "health-check",
    "url": "https://api.example.com/health",
    "kind": "recurring",
    "every_n_minutes": 5
  }
}

pause_job

Pause a running job. It will stop firing until resumed.

resume_job

Resume a paused job. TrigRun recomputes the next run time from now.

run_job

Trigger immediate execution of a job, regardless of its schedule.

{
  "tool": "run_job",
  "arguments": {
    "token": "cron_pat_abc123...",
    "job_id": "clx9abc123def"
  }
}

delete_job

Permanently delete a job and its execution history.

Executions

list_executions

List executions. Optionally filter to a specific job.

{
  "tool": "list_executions",
  "arguments": {
    "token": "cron_pat_abc123...",
    "job_id": "clx9abc123def"
  }
}

get_execution

Get full details for an execution, including all retry attempts and the final response.

{
  "tool": "get_execution",
  "arguments": {
    "token": "cron_pat_abc123...",
    "execution_id": "clx9exec456ghi"
  }
}

replay_execution

Re-queue a specific execution. Useful for recovering failed runs without waiting for the next scheduled fire.

Notification channels

list_channels

List all notification channels in your workspace.

create_channel

Create a notification channel. Supported types: webhook, email, slack, discord.

{
  "tool": "create_channel",
  "arguments": {
    "token": "cron_pat_abc123...",
    "name": "ops-alerts",
    "type": "slack",
    "url": "https://hooks.slack.com/services/T00000000/B00000000/XXXXXXXXXXXXXXXXXXXXXXXX"
  }
}

delete_channel

Delete a notification channel.

Notification rules

list_rules

List notification rules attached to a job.

add_rule

Attach a notification channel to a job event.

{
  "tool": "add_rule",
  "arguments": {
    "token": "cron_pat_abc123...",
    "job_id": "clx9abc123def",
    "channel_id": "clx9ch789jkl",
    "event": "on_failure"
  }
}

remove_rule

Remove a notification rule from a job.

Secrets

list_secrets

List all secret names in your workspace. Values are never returned.

create_secret

Store an encrypted secret. Reference it in job headers with secret://name.

{
  "tool": "create_secret",
  "arguments": {
    "token": "cron_pat_abc123...",
    "name": "my-api-key",
    "value": "sk-live-abc123..."
  }
}

To use this secret in a job header:

{
  "headers": { "Authorization": "Bearer secret://my-api-key" }
}

delete_secret

Delete a secret permanently.

Notification deliveries

list_deliveries

List notification deliveries. Optionally filter by status.

retry_delivery

Retry a failed notification delivery.

Common workflows

Create a cron job with failure alerts

This is the most common setup: a regularly scheduled job with a Slack alert when it fails.

1. Create a Slack channel:

   "Create a Slack notification channel named 'cron-alerts' with webhook URL https://hooks.slack.com/... using token cron_pat_..."

2. Create the job:

   "Create a cron job named 'daily-report' that POSTs to https://api.example.com/report every day at 8 AM UTC using token cron_pat_..."

3. Attach the alert:

   "Add a notification rule to job JOB_ID: send to channel CHANNEL_ID on_failure using token cron_pat_..."

The agent calls create_channel, create_job, then add_rule in sequence.

Check execution status

After a job fires, inspect what happened:

1. List recent executions:

   "List executions for job JOB_ID using token cron_pat_..."

2. Get details on a specific one:

   "Get execution EXEC_ID using token cron_pat_..."

   The response includes every retry attempt, HTTP status codes, latency, and the full response body.

3. If it failed, check why: The get_execution result shows attempts with status, statusCode, and responseBody for each attempt.

Replay a failed execution

When an execution failed due to a transient error (timeout, 503), replay it without waiting for the next scheduled run:

1. Find the failed execution:

   "List executions for job JOB_ID using token cron_pat_..."

2. Replay it:

   "Replay execution EXEC_ID using token cron_pat_..."

   replay_execution re-queues the exact same request. TrigRun will apply the job's retry policy to the replay as well.

Store a secret and use it in a job

Avoid hardcoding API keys in job headers:

1. Store the secret:

   "Create a secret named 'stripe-api-key' with value sk_live_... using token cron_pat_..."

2. Create the job referencing the secret:

   "Create a job named 'process-payments' that POSTs to https://api.example.com/payments with header Authorization: Bearer secret://stripe-api-key, using token cron_pat_..."

   The worker resolves secret://stripe-api-key at execution time. The raw value is never stored in the job record.

Temporarily pause and resume a job

To disable a job without deleting it (e.g., during a maintenance window):

1. Pause:

   "Pause job JOB_ID using token cron_pat_..."

2. Resume when ready:

   "Resume job JOB_ID using token cron_pat_..."

   Resuming recomputes the next run from the current time — it won't backfill missed fires.

Tool reference summary