feat: support code deploy in --no-prompt mode via CLI flags

[v1212]

Summary

Adds --deploy-mode, --runtime, --entry-point, and --dep-resolution flags to azd ai agent init for non-interactive code deploy support in CI/CD pipelines.

Closes #8323

What this PR changes

In --no-prompt mode, the deploy mode was hardcoded to "container" with no way to select code deploy. This PR adds four new flags:

Flag	Values	Default (no-prompt)	Required
--deploy-mode	container, code	container	No
--runtime	python_3_13, python_3_14, dotnet_10	ΓÇö	Yes, when --deploy-mode code --no-prompt
--entry-point	file path (e.g. app.py)	ΓÇö	Yes, when --deploy-mode code --no-prompt
--dep-resolution	remote_build, bundled	remote_build	No

Example usage

azd ai agent init --no-prompt \
  --project-id "/subscriptions/.../projects/my-project" \
  --model-deployment "gpt-4o" \
  --deploy-mode code \
  --runtime python_3_13 \
  --entry-point app.py

Interaction with existing --no-prompt behavior

This PR does not change how --no-prompt handles Foundry project or model deployment selection. The existing behavior remains:

Decision	--no-prompt behavior (unchanged)
Foundry project	--project-id is required (must be an existing project). Without it, init errors out with a hint to pass --project-id.
Model deployment	If --model-deployment is passed, looks up that deployment in the Foundry project. If not passed, auto-selects the first matching deployment. If none exist, auto-configures a new deployment from the catalog (requires azd up).
End-of-init message	Existing deployment selected → azd deploy. New deployment created → azd up (provision needed to create the model deployment via Bicep).

The new --deploy-mode / --runtime / --entry-point flags only control how the agent code is packaged and deployed (container image vs ZIP upload). They are orthogonal to project and model selection.

Tests added

• TestPromptDeployMode_FlagOverride ΓÇö 6 cases: flag precedence, invalid flag error, noPrompt default
• TestPromptCodeConfig_FlagOverrides ΓÇö 5 cases: opts override, noPrompt defaults for Python/dotnet
• TestCodeDeployFlagValidation ΓÇö 5 cases: validates required flags in noPrompt+code mode

---

E2E Test Results

Build: azd x build --cwd cli/azd/extensions/azure.ai.agents on branch feat/no-prompt-code-deploy
Extension: 0.1.33-preview (local dev build)
Project: Existing Foundry project in North Central US with gpt-4.1-mini deployment

Test Matrix

#	Scenario	Command Flags	Result
1	Full code deploy (init + deploy + invoke)	--deploy-mode code --runtime python_3_13 --entry-point main.py --dep-resolution remote_build	PASS
2	Bundled dependency resolution	--deploy-mode code --runtime python_3_13 --entry-point main.py --dep-resolution bundled	PASS
3	Missing --runtime validation	--deploy-mode code --entry-point main.py	PASS (error)
4	Missing --entry-point validation	--deploy-mode code --runtime python_3_13	PASS (error)
5	Invalid --deploy-mode validation	--deploy-mode invalid	PASS (error)
6	Backward compat (no --deploy-mode)	(no deploy-mode flag)	PASS (defaults to container)
7	Explicit container mode	--deploy-mode container	PASS

Key Verifications

• Case 1: azd ai agent init --no-prompt with all flags agent.yaml contains correct code_configuration (runtime, entry_point, dependency_resolution) azd deploy succeeds agent invoke returns valid LLM response
• Cases 3-5: Proper validation errors with helpful suggestions when required flags are missing
• Cases 6-7: Default behavior unchanged no --deploy-mode flag still defaults to container (backward compatible)

All tests used --no-prompt mode with -m <manifestUrl> and --project-id.

[Copilot]

Pull request overview

Adds non-interactive (“--no-prompt”) support for code deploy in azd ai agent init by introducing CLI flags that allow CI/CD pipelines to fully specify deploy mode and code-deploy configuration without prompts.

Changes:

• Adds --deploy-mode, --runtime, --entry-point, and --dep-resolution flags to azd ai agent init.
• Implements shared validation for required flags in --no-prompt + --deploy-mode code flows and threads flag overrides into the existing prompt helpers.
• Adds unit tests covering deploy-mode flag precedence, code config flag overrides, and required-flag validation.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
cli/azd/extensions/azure.ai.agents/internal/cmd/init.go	Registers new CLI flags, validates required inputs for non-interactive code deploy, updates help/examples, and passes flag overrides into prompting.
cli/azd/extensions/azure.ai.agents/internal/cmd/init_from_code.go	Updates deploy-mode and code-config prompt helpers to accept flag overrides and validates required flags in the init-from-code flow.
cli/azd/extensions/azure.ai.agents/internal/cmd/init_test.go	Adds tests for required-flag validation behavior in --no-prompt + code deploy mode.
cli/azd/extensions/azure.ai.agents/internal/cmd/init_from_code_test.go	Adds tests for deploy-mode flag precedence and code-config flag override behavior.

Comments suppressed due to low confidence (1)

cli/azd/extensions/azure.ai.agents/internal/cmd/init_from_code.go:1292

• When opts.runtime is provided (via --runtime), it’s accepted verbatim without validation against the supported runtime values shown in the prompt/help text. This can produce an agent.yaml with an invalid runtime and later failures (e.g., wrong command prefix). Add validation for opts.runtime (and ideally for interactive flag usage too) to ensure it matches one of the supported values (python_3_13/python_3_14/dotnet_10) and return a structured validation error when invalid.

	var runtime string
	if opts.runtime != "" {
		runtime = opts.runtime
	} else if noPrompt {
		if isDotnet && !isPython {
			runtime = "dotnet_10"
		} else {
			runtime = "python_3_13" // default to Python for mixed/unknown repos (language preference, not version compat)

[trangevi]

/check-enforcer override