Draft: prototype for init/status, dry-run, styled help, and json output

[ikotelnikov]

This is a draft prototype PR to show a possible direction for several related TRCLI improvements:

• init command
• status command
• styled / colorized help output
• --dry-run support
• --json support for structured output where applicable

The goal is to make TRCLI easier to set up, safer to automate, and easier to integrate with CI / AI-assisted workflows.

This PR is intentionally shared as a prototype first to get feedback on direction and scope.
If the approach looks useful, I can split this into smaller follow-up PRs for review.

[acuanico-tr-galt]

Hi @ikotelnikov thank you for opening this PR! its currently under review, we'll get back to you soon. If you don't mind what specific use cases do you intend to utilize these features for? Thank you.

[ikotelnikov]

Hi @acuanico-tr-galt thanks for answering!

The main goal of this PR is to show a prototype. I shared it with the TestRail Product Manager via email. I’m not really proposing to review and merge this code as-is. If any of the ideas are interesting, it would make more sense to split them into smaller, separate PRs.

The features here fall into two directions:

1. init, status, --help (with examples and trcli official docs linked) + ANSI colors → improve the first-time experience when setting up trcli in a new environment.

2. --dry-run and --json → improve usage with agents and automation. I don’t have specific production use cases here — this is based on two observations:

• Codex CLI tends to use --dry-run when available during the installation process of new tools.
• JSON output in stdout is more reliable and standardized for automation workflows. At the moment (as trcli is mostly focused on test result ingestion), the value of --json is limited to exposing integration artifacts not available beforehand (e.g., run_id, auto-created cases, sync errors), but as the interface expands (commands to read cases, runs, etc. were proposed in an issue), standardized machine-readable output becomes valuable.

Overall, you can just run the prototype, see how it works in practice, and feel free to close this PR.