Agent Skills: zapier

Zapier API for workflow automation. Use when user mentions "Zapier",

UncategorizedID: vm0-ai/vm0-skills/zapier

Install this agent skill to your local

pnpm dlx add-skill https://github.com/vm0-ai/vm0-skills/tree/HEAD/zapier

Skill Files

Browse the full folder contents for zapier.

Download Skill

Loading file tree…

zapier/SKILL.md

Skill Metadata

Name
zapier
Description
Zapier API for workflow automation. Use when user mentions "Zapier",

Troubleshooting

If requests fail, run zero doctor check-connector --env-name ZAPIER_TOKEN or zero doctor check-connector --url https://actions.zapier.com/api/v2/check/ --method GET

How to Use

All examples below assume you have ZAPIER_TOKEN set. Authentication uses the x-api-key header.

1. Check API Key

Verify that your API key is valid.

curl -s "https://actions.zapier.com/api/v2/check/" --header "x-api-key: $ZAPIER_TOKEN" | jq .

2. List Exposed Actions

Retrieve all actions you have configured and exposed in your Zapier AI Actions dashboard.

curl -s "https://actions.zapier.com/api/v1/exposed/" --header "x-api-key: $ZAPIER_TOKEN" | jq '.results[] | {id, description, params}'

3. Execute an AI Action

Execute a configured action using natural language instructions. Replace ACTION_ID with the action ID from the list above.

curl -s -X POST "https://actions.zapier.com/api/v2/ai-actions/ACTION_ID/execute/" --header "Content-Type: application/json" --header "x-api-key: $ZAPIER_TOKEN" -d '{"instructions": "Send a message saying hello to the #general channel"}' | jq .

4. Execute with Parameter Hints

Supply parameter hints to guide the AI in filling action fields. Supported modes: locked (use exact value), guess (AI matches text), choose_from (AI selects from list), ignored (skip parameter).

Write to /tmp/zapier_request.json:

{
  "instructions": "Send an email about the weekly report",
  "params": {
    "to": {
      "mode": "locked",
      "value": "team@example.com"
    },
    "subject": {
      "mode": "locked",
      "value": "Weekly Report"
    }
  }
}

Then run:

curl -s -X POST "https://actions.zapier.com/api/v2/ai-actions/ACTION_ID/execute/" --header "Content-Type: application/json" --header "x-api-key: $ZAPIER_TOKEN" -d @/tmp/zapier_request.json | jq .

5. Preview an Action (Dry Run)

Preview what the action would do without actually executing it. Add preview_only=true as a query parameter.

curl -s -X POST "https://actions.zapier.com/api/v2/ai-actions/ACTION_ID/execute/?preview_only=true" --header "Content-Type: application/json" --header "x-api-key: $ZAPIER_TOKEN" -d '{"instructions": "Create a new row in the Sales spreadsheet with name John and amount 500"}' | jq .

6. Execute a V1 Exposed Action

Execute an action using the V1 endpoint. Replace ACTION_ID with the exposed action ID.

curl -s -X POST "https://actions.zapier.com/api/v1/dynamic/exposed/ACTION_ID/execute/" --header "Content-Type: application/json" --header "x-api-key: $ZAPIER_TOKEN" -d '{"instructions": "Send a Slack message to #dev saying deployment complete"}' | jq .

Guidelines

  1. Configure actions first: Before using the API, you must configure and enable actions in the Zapier AI Actions dashboard. The API can only execute actions you have set up
  2. Use natural language: The instructions field accepts plain English descriptions of what you want to do. The AI interprets and maps instructions to the correct action parameters
  3. Parameter modes: Use locked mode when you know the exact value, guess when the AI should match text to available options, and choose_from to provide a list of valid IDs
  4. Preview before executing: Use preview_only=true to verify the AI correctly interprets your instructions before running the action
  5. Response statuses: Check the status field in responses - values are success, error, empty, or preview
  6. Rate limits: Zapier enforces rate limits on API calls. Implement backoff if you receive HTTP 429 responses
  7. Action IDs: Each configured action has a unique ID. Use the list endpoint to discover available action IDs