screener-workflow

name	screener-workflow
description	Operate TradingView Stock Screener with the Rust `tv` CLI. Use when the user wants Screener reads, scanner-to-screener workflow, saved screen switching, test screen setup, filter or column changes, or cleanup of disposable Screener state.

Screener Workflow

Use this skill for TradingView Stock Screener work through the Rust tv CLI. Screener work spans two source categories: tv scanner ... commands are Desktop-free reads, while tv screener ... commands are Desktop-backed reads or operations against the visible/full-page TradingView Screener target.

For follow-up chart observation after a screen, use docs/observation-workflows.md rather than expanding this skill with chart analysis steps.

Start With The Target

1. Run tv readiness for Desktop/CDP health and chart target ambiguity.
2. Run tv tab list and prefer a full-page Screener target from screener_targets. Preserve source_category, requires_desktop, and non_mutating when reporting Desktop-backed target evidence.
3. Use the returned target_cli_args, for example tv --target-id <ID> screener ..., for follow-up commands.
4. If no full-page Screener target is available, run tv screener open --full-page and then use the returned target_cli_args.

Read First

Start with read-only commands before changing saved state:

• tv screener status
• tv screener get --limit <N>
• tv screener screens active
• tv screener screens list
• tv screener screens list --catalog
• tv screener filters list
• tv screener filters actions
• tv screener columns list
• tv screener columns config
• tv screener columns actions

Use tv scanner hotlist and tv scanner scan for broad Desktop-free market discovery. Preserve their source_category: "desktop_free_read", requires_desktop: false, and non_mutating: true metadata when comparing them with Desktop-backed Screener rows. Use tv screener get when the user needs the currently visible Screener rows. Use tv scanner metainfo --field <FIELD> when you need to confirm whether a scanner field exists before building a scan. If visible Screener state remains unclear after structured reads, use tv screenshot --region full --output <PATH> as portable visual evidence. Screenshots do not mutate TradingView state, but they do write a local file.

Interpreting Results

For result explanation, use screener-result-analysis. Keep this skill focused on operating the Screener target and saved state. A compact first pass is usually enough: source, filters or screen name, columns, sort key, result count, top rows, and any intended next read.

Mutations

Use --dry-run first whenever the command supports it. Explain the expected saved-state change and get user approval before normal mutation.

• Screen switching: tv screener screens switch --name <NAME> [--catalog] --dry-run.
• Screen lifecycle: use disposable names containing CLI-Test or テスト for create, rename, save-as, and delete.
• Screen save: use only on prepared test or disposable screens.
• Filters: use filters add, modify, remove, and clear with dry-run target checks first.
• Columns: use columns config before columns add, remove, or reorder. columns add requires a known storage column id and optional JSON-object params; it is not display-name search.

Boundaries

• columns reset is deferred because no reliable default source is confirmed.
• Broad multi-option and free-text filter editing are not genericized.
• Normal screen lifecycle, storage-backed filter cleanup, and column mutations should stay on test or disposable screens unless the user explicitly accepts account-state changes.
• Do not record real screen ids, account-local names, raw storage payloads, or target ids in shared notes.

Reporting

Report which target was used, which commands were read-only or dry-run, what mutated, and whether post-checks confirmed the requested after-state. If any test screen, filter, or column remains changed, name only the visible disposable object and the intended cleanup command.