[szmyty]

Mindcap currently mixes progress with result output, prints raw exception text, repeats every failed asset warning, and can count a partial sync archive as a successful capture. This gives CLI commands consistent timestamped diagnostics, safe error handling, and usable failure summaries.

• Add bright severity labels, literal Rich text, grouped repeated warnings, attempted-item counts and elapsed time. Keep final results on stdout and progress/warnings/errors on stderr, including export imports and interrupted sync JSON. Preserve full archive paths at narrow widths.
• Add global --verbose, --debug, --quiet and optional --log-file (before the command name), preserving capture/sync's existing local flags. JSONL logs append with owner-only permissions; diagnostic sessions restore handlers and close files after each command. Debug failures show sanitized frame locations without source lines or locals. Root/HTTP loggers remain untouched.
• Give operational and argument errors stable codes and next steps. Redact common credential fields, named session cookies, nested cookie/storage-state objects, JWTs, URL credentials/queries/fragments, and terminal control sequences. Do not log provider payloads or browser state.
• Return exit 2 for incomplete single captures and explicit incomplete batch items while retaining saved archives and their integrity results. Display retries and final failures, retain safe warnings/errors in sync reports, reconcile identical warnings separately for each source, and restore signal handlers so interrupted runs remain interrupted and resumable (exit 130).

Dependency: builds on Suno fix #38. Merge #38 first, then this PR. The PR targets main so it needs no manual base change. Logging-only comparison. Bookmark #37 remains separate and unmerged.

Validation on exact tree 384c432f88de98da6049affb328697551299a74c:

• 611 tests passed, including architecture boundaries and synthetic interruption, retry, warning grouping, JSON channels, file permissions, handler cleanup, and credential redaction.
• Ruff lint/format passed with both the uv-lock and Poetry-lock Ruff versions; focused strict Mypy passed 10 changed source/test files. Full-repository Mypy retains exactly the same 28 existing diagnostics in nine files.
• Wheel and sdist build. A fresh wheel-only install outside the checkout passed nested-command logging, an offline ChatGPT fixture capture, JSON errors, redacted argument errors and JSONL logging using newly resolved Typer 0.27.3; locked tests use 0.27.0.
• Inspected a synthetic colored terminal preview, including a 36-audio-failure summary and actionable error panel.

No real credentials, Chrome profiles, source accounts or media downloads were accessed. Diagnostics can still include identifiers, titles and paths; they are not anonymous exports. Unrecognized-option usage errors may render text even when --json is present; operational JSON results and errors use their documented channels. The cause of the maintainer's live Suno 403s still needs the diagnostic result requested in #38. No runtime dependencies or lockfiles change.

GitHub Actions CI run 39132605708 passed on published head 5e818da7069c7891f29c5aa14c260acf1d36d0f2: dependency install/configuration, formatting, lint, full tests, architecture boundaries, wheel/sdist build, package-content validation and clean-tree checks. The published tree matches the locally tested tree exactly.

[devactivity-app]

Pull Request Summary by devActivity

Metrics

Achievements

• @szmyty
  Earned XP: 15⭐
  Check Your Progress