fix: treat task --help and unknown flags as CLI errors (#539)

[stantheman0128]

Summary

• parseArgs gains rejectUnknownOptions so unrecognized --flags throw instead of becoming prompt text.
• Subcommand parsers enable that mode, accept --help/-h, and print usage without starting a Codex thread.
• Fixes task --help (and any unknown --flag) is swallowed into the prompt and dispatches a real thread, hijacking --resume-last #539 (Windows-reproable; defect is in JS arg handling).

AI assistance

Assisted by Cursor/Grok. I reviewed the diff, ran the tests below, and checked open PRs for overlap (#128/#535 are test-only coverage; #378 is adversarial focus parsing).

Evidence

node --test tests/args.test.mjs tests/commands.test.mjs tests/render.test.mjs
# 14 pass (includes task --help + unknown --flag with fake-codex: no threads)

node --test --test-name-pattern "task runs when the active provider" tests/runtime.test.mjs
# 1 pass

Test plan

• task --help --cwd <tmp> --json prints Usage and exits 0
• task --not-a-real-flag errors with Unknown option and does not create a thread
• Normal task prompt still runs under fake-codex

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0e0cd92549

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Keep short-flag text usable in task prompts

When task is invoked in the documented quoted raw-argument form, normalizeArgv re-splits the natural-language prompt before it reaches this parser, so any prompt that mentions a short CLI flag or negative-style token now fails here before Codex is dispatched. For example, task "investigate grep -R usage" exits with Unknown option: -R instead of sending that prompt, which is a common shape for debugging command-line behavior; the parser should either stop option parsing once prompt text starts or avoid rejecting single-dash tokens for prompt-taking commands.

Useful? React with 👍 / 👎.