Code Suggestions API
Use the Code Suggestions API to access the Code Suggestions feature.
Create an access token
- Introduced in GitLab 16.1.
Creates an access token to access Code Suggestions.
POST /code_suggestions/tokens
curl --request POST \
--header "PRIVATE-TOKEN: <your_access_token>" \
--url "https://gitlab.example.com/api/v4/code_suggestions/tokens"
Example response:
{
"access_token": "secret-access-token",
"expires_in": 3600,
"created_at": 1687865199
}
Generate code completions
DETAILS: Status: Experiment
- Introduced in GitLab 16.2 with a flag named
code_suggestions_completion_api
. Disabled by default. This feature is an Experiment.- Requirement to generate a JWT before calling this endpoint was removed in GitLab 16.3.
- Generally available in GitLab 16.8. Feature flag
code_suggestions_completion_api
removed.
FLAG:
On self-managed GitLab, by default this feature is not available. To make it available, an administrator can enable the feature flag named code_suggestions_completion_api
.
On GitLab.com, this feature is not available.
This feature is not ready for production use.
NOTE: This endpoint rate-limits each user to 60 requests per 1-minute window.
Use the AI abstraction layer to generate code completions.
POST /code_suggestions/completions
Requests to this endpoint are proxied directly to the model gateway. The documentation for the endpoint is currently the SSoT for named parameters.
curl --request POST \
--header "Authorization: Bearer <YOUR_ACCESS_TOKEN>" \
--data "<JSON_BODY>" \
--url "https://gitlab.example.com/api/v4/code_suggestions/completions"
Example body:
The model gateway is the SSoT for parameters.
Example response:
{
"id": "id",
"model": {
"engine": "vertex-ai",
"name": "code-gecko"
},
"object": "text_completion",
"created": 1688557841,
"choices": [
{
"text": "\n if self.is_running:\n self.speed += increment\n print(\"The car's speed is now",
"index": 0,
"finish_reason": "length"
}
]
}