Pull request overview

Updates the flash init CLI behavior so invoking it without a project argument shows a usage/help panel and exits, making “initialize in current directory” an explicit flash init ..

Changes:

• Added an early-exit usage panel for flash init when called with no positional argument.
• Updated init logic so only flash init . initializes the current directory (no longer the default when omitted).
• Added/updated unit tests to cover the new no-args behavior (exit code, no skeleton creation, usage output).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Code Review — PR #235: flash init usage menu

The init.py change is clean — showing usage panel when no args instead of silently initializing in cwd is good UX.

Note: This PR shares a large diff overlap with PRs #221 and #237 (aiohttp session consolidation, api_key_context.py removal, DEFAULT_WORKERS_MIN change, get_api_key() migration). The bugs found in those shared components apply here too:

1. HIGH — Authorization header sent to presigned S3 URLs in app.py (see PR #221 review)
2. HIGH — Per-request API key propagation from LB to workers silently removed (see PR #221 review)
3. HIGH — uv pip install fails without active venv in update.py (see PR #237 review, if update.py is included)

The init.py change itself has no issues. Tests cover all three modes (flash init, flash init ., flash init name).

just that one fix and then should be good to go

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.

1. Implementation — behavior change

Clean change. The None branch was cleanly split from "." — previously if project_name is None or project_name == "." routed both to CWD init; now only "." does. The ctx.get_help() approach is the right Typer idiom: it surfaces the same text as flash init --help, is automatically kept in sync as the command signature changes, and the markup=False, highlight=False flags prevent Rich from misinterpreting brackets in the output as markup.

raise typer.Exit(0) is correct — showing help is not an error, so exit code 0 is right.

2. Tests

Question: no CLI-level test for Typer context injection

All tests call init_command(mock_typer_ctx, ...) directly, bypassing Typer's wiring. The unit tests can't verify that ctx: typer.Context is correctly recognized and injected when invoked as a real CLI command. If something were misconfigured at the app.command() registration level, the unit suite would stay green while the CLI silently broke.

Adding a single CliRunner test covers this:

from typer.testing import CliRunner
from runpod_flash.cli.main import app

def test_no_args_via_cli():
    result = CliRunner().invoke(app, ["init"])
    assert result.exit_code == 0
    assert "flash init" in result.output

This is low risk given that typer.Context injection is well-established, but it's the one gap the current tests leave open.

The three new TestInitCommandNoArgs tests cover the right cases — exit code, no skeleton created, help printed with correct kwargs.

Deleted test (test_init_current_directory_with_none) — correct to remove. Its replacement is TestInitCommandNoArgs, which is clearer.

Nits

• test_no_args_prints_usage_info checks "flash init" in help_arg but the string comes from the mock fixture ("Usage: flash init [OPTIONS] [PROJECT_NAME]"). The test doesn't explicitly assert that ctx.get_help() was called — mock_typer_ctx.get_help.assert_called_once() would make the intent clearer.
• The raw --help output is functional but a custom panel with two concrete examples (flash init . and flash init my-project) would be more discoverable for users who don't know the dot convention. Enhancement only — not blocking.

Verdict: PASS WITH NITS

The behavior change is clean and correct. The one meaningful gap is a CLI-level integration test to confirm Typer's context injection. The nits are minor.

🤖 Reviewed by Henrik's AI-Powered Bug Finder

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.