- Mark help parser with `_is_help_command=True` so CLI renders as `program help`.
- Add TLDR examples to `Exit` and `History` commands.
- Normalize help TLDR/tag docs to short forms `-T` (tldr) and `-t [TAG]`.
- Also propagate submenu exit help text TLDRs when set.
- Disallow defaults for `HELP`, `TLDR`, `COUNT`, and boolean store actions.
- Enforce list defaults for `APPEND`/`EXTEND` and any `nargs` in `{int, "*", "+"}`; coerce to list when `nargs == 1`.
- Validate default(s) against `choices` (lists must be subset).
- Strengthen `choices` checking at parse-time for both scalars and lists; track invalid-choice state for UX.
- New `_resolve_posix_bundling()` with context:
- Won’t split negative numbers or dash-prefixed positional/path values.
- Uses the *last seen flag’s type/action* to decide if a dash token is a value vs. bundle.
- Add `_is_valid_dash_token_positional_value()` and `_find_last_flag_argument()` helpers.
- Completions overhaul
- Track `consumed_position` and `has_invalid_choice` per-arg (via new `ArgumentState.set_consumed()` / `reset()`).
- Add `_is_mid_value()` and `_value_suggestions_for_arg()` to produce value suggestions while typing.
- Persist value context for multi-value args (`nargs="*"`, `"+"`) for each call to parse_args
- Suppress suggestions when a choice is currently invalid, then recover as the prefix becomes valid.
- Respect `cursor_at_end_of_token`; do not mutate the user’s prefix; improve path suggestions (`"."` vs prefix).
- Better behavior after a space: suggest remaining flags when appropriate.
- Consistent `index` naming (vs `i`) and propagate `base_index` into positional consumption to mark positions accurately.
- Return value tweaks for `find_argument_by_dest()` and minor readability changes.
- Replace the minimal completion test with a comprehensive suite covering:
- Basics (defaults, option parsing, lists, booleans).
- Validation edges (default/choices, `nargs` list requirements).
- POSIX bundling (flags only; negative values; dash-prefixed paths).
- Completions for flags/values/mid-value/path/`nargs="*"` persistence.
- `store_bool_optional` (feature / no-feature, last one wins).
- Invalid choice suppression & recovery.
- Repeated keywords (last one wins) and completion context follows the last.
- File-system-backed path suggestions.
- Bumped version to 0.1.83.
- Updated `Command.help_signature` to return a `(Panel, description)` tuple,
enabling richer Rich-based help output with formatted panels.
- Integrated `program` context into commands to display accurate CLI invocation
(`falyx run …`) depending on mode (RUN, PREVIEW, RUN_ALL).
- Refactored `_show_help` to print `Panel`-styled usage and descriptions instead
of table rows.
- Added `program` and `options_manager` propagation to built-in commands
(Exit, History, Help) for consistent CLI display.
- Improved `CommandArgumentParser.add_tldr_examples()` with stricter validation
(`all()` instead of `any()`), and added new TLDR tests for coverage.
- Simplified parser epilog text to `Tip: Use 'falyx run ?' to show available commands.`
- Added tests for required `Argument` fields and TLDR examples.
- Bumped version to 0.1.73.
- Introduce `ArgumentAction.TLDR` for showing concise usage examples
- Add `rich_text_to_prompt_text()` to support Rich-style markup in all prompt_toolkit inputs
- Migrate all prompt-based Actions to use `prompt_message` with Rich styling support
- Standardize `CancelSignal` as the default interrupt behavior for prompt-driven Actions
