gcp-vertexai-generate-content-input
Generative model prompts, system instructions, function responses, and tool configuration can transmit plaintext credentials or sensitive data.
Where it sits
| location | aiplatform.projects.locations.publishers.models.generateContent.request.{contents[].parts[],systemInstruction.parts[],tools[]} |
|---|
| location kind | payload |
|---|
| data kinds | credential password api_key oauth_token customer_data pii sensitive_data |
|---|
| emits edge | ContainsCredential |
|---|
| service | Vertex AI / Workbench (gcp:vertexai) |
|---|
Collection recipe
| access mode | write_only_input |
|---|
| operation | aiplatform.projects.locations.publishers.models.generateContent |
|---|
| response path | request.{contents[].parts[],systemInstruction.parts[],tools[]} |
|---|
| encoding | json |
|---|
| params | {"name": "\u003cresource-name\u003e"} |
|---|
Required permissions
aiplatform.endpoints.predict
References