Datasets:
roanbrasil
/
devops-gitops-corpus

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
0
59.1k
```ts
import { createTool } from "@voltagent/core";
import { VoltOpsClient } from "@voltagent/core";
import { z } from "zod";
const voltops = new VoltOpsClient({
publicKey: process.env.VOLTAGENT_PUBLIC_KEY!,
secretKey: process.env.VOLTAGENT_SECRET_KEY!,
});
export const createAirtableRecordTool = createTool({
name: "createAirtableRecord",
description: "Create a CRM row inside Airtable",
parameters: z.object({
fields: z.record(z.unknown()),
baseId: z.string().optional(),
tableId: z.string().optional(),
}),
execute: async ({ fields, baseId, tableId }) => {
const result = await voltops.actions.airtable.createRecord({
credential: { credentialId: process.env.AIRTABLE_CREDENTIAL_ID! },
baseId: baseId ?? process.env.AIRTABLE_BASE_ID!,
tableId: tableId ?? process.env.AIRTABLE_TABLE_ID!,
fields,
typecast: true,
});
return {
actionId: result.actionId,
metadata: result.metadata,
record: result.responsePayload,
};
},
});
```
Add the tool to an agent:
```ts
const airtableAgent = new Agent({
name: "Airtable Assistant",
model: openai("gpt-4o-mini"),
instructions: "Use the Airtable tools to manage the operations database.",
tools: [createAirtableRecordTool, listAirtableRecordsTool, updateAirtableRecordTool],
});
```
Now the agent can plan/tool-call requests like:
```
User: "Add Ada Lovelace to the workspace table and show me the latest 5 entries."
```
The planner will call `createAirtableRecord` followed by `listAirtableRecords`, and VoltOps will log
both action runs.
## Testing payloads in the console
While building, open **Volt Console → Actions → Add Action** and select Airtable. The drawer mirrors
the trigger UX with a credential step, configuration step (base/table dropdowns), and a **Payload &
Test** editor. Enter any JSON payload and click **Run Test**; successful responses include a
copy-pasteable SDK snippet, while errors display the raw provider message (e.g.,
`INVALID_VALUE_FOR_COLUMN`).
This makes it easy to iterate on Airtable data types before committing changes to an agent.
## Troubleshooting
- **Proxy/credential mismatch** – make sure the Volt credential uses the same service (case
sensitive). E.g., a credential named “airtable” must be attached to an Airtable action.
- **Column IDs vs names** – use `returnFieldsByFieldId: true` if you prefer Airtable field IDs; the
default returns user-friendly names.
- **Pagination** – `listRecords` accepts `pageSize`, `offset`, `view`, `filterByFormula`, and `sort`
parameters identical to Airtable’s REST API. VoltOps just forwards them.
- **Observability** – check **Volt → Actions → Runs** to inspect payloads, metadata, retries, and
provider errors (`INVALID_PERMISSIONS_OR_MODEL_NOT_FOUND`, etc.).
With VoltOps Actions you can keep agent logic focused on reasoning while Volt handles the last-mile
integration to Airtable.
<|endoftext|>
# source: VoltAgent__voltagent/website/actions-triggers-docs/actions/overview.md type: docs
---
title: Actions Overview
---
# Actions Overview
VoltOps Actions provide managed integrations (Airtable, Gmail, Slack, …) that you can trigger from
agents, workflows, or the VoltOps API. Instead of wiring every SaaS API yourself you define the
destination once in the Volt console, connect credentials, and then call the action from the SDK or
VoltAgent tools with full observability.
## When to use actions
- **Deliver data to external systems** – sync agent output to Airtable, send emails, notify chat apps,
or call internal HTTP endpoints.
- **Re-use credentials** – end users authenticate once inside Volt; your code invokes the action
without ever handling another API token.