schema_version = 1

extends     = "OpenAI Base Chat"
name        = "OpenAI Completion"
description = "OpenAI GPT-5.4 mini — code completion using the <code_context> chat format over Chat Completions. reasoning_effort=none for low latency. GPT-5 models reject 'temperature' AND 'stop' on Chat Completions, so completion cannot be pinned to t=0 and has no stop sequence — it relies on the prompt + max_completion_tokens to stay short."

model = "gpt-5.4-mini"
tags  = ["completion", "openai", "cloud"]

system_prompt = """
{%- if language == "qml" %}{{ read_file(":/roles/code-completion-qml.md") }}
{%- else if language == "c-like" %}{{ read_file(":/roles/code-completion-c-like.md") }}
{%- else %}{{ read_file(":/roles/code-completion.md") }}
{%- endif %}
{{ read_file(":/tasks/code-completion.md") }}"""

[body]
max_completion_tokens = 1024
reasoning_effort      = "none"
messages = """
[
  {% if existsIn(ctx, "system_prompt") %}
  { "role": "system", "content": {{ tojson(ctx.system_prompt) }} },
  {% endif %}
  {
    "role": "user",
    "content": {{ tojson("Here is the code context with insertion points:\\n<code_context>\\n" + ctx.prefix + "<cursor>" + ctx.suffix + "\\n</code_context>") }}
  }
]
"""