ZenTable Skill

Render structured table data into high-quality PNG images.

TL;DR

ZenTable turns messy table-like content into readable, decision-ready table outputs for both mobile (Discord-first) and desktop.

Accepted inputs

• - text-based content (raw text tables, long agent responses)
• structured JSON
• screenshots / real-world photos (via OCR-assisted extraction when needed)

Core capabilities

• - CSS + PIL rendering
• sorting / filtering / pagination
• threshold-based highlighting
• optional PNG + TXT dual output (--both)

Runtime & security note

This skill runs local scripts and depends on local runtime binaries (python3, google-chrome).
Review code and dependencies before running in sensitive environments.

Naming convention

• - Canonical code name: zentable (lowercase)
• UI / brand label: INLINECODE4
• INLINECODE5 / zenble* are legacy compatibility aliases
• Reference: NAMING_MIGRATION.md (repository document)

When to use

✅ Use this skill when:

• - You need a visual table image instead of plain text
• You need polished output for chat/report/social sharing
• The dataset is large enough that plain text is hard to read
• You need a specific visual theme (iOS-like, dark, compact, etc.)

❌ Do not use this skill when:

• - The table is tiny and plain text is enough
• The user explicitly asks for no image output
• The user needs an editable spreadsheet format (CSV/Excel)

Capability matrix (SkillHub release)

Known limitations

• - ASCII output is sensitive to platform font fallback and whitespace handling.
• Create separate calibration profiles per platform; do not share blindly.
• INLINECODE9 already includes text-theme fallback to default when no text theme exists.
• Discord plain text collapses repeated normal spaces; Unicode spacing characters may be required for spacing preservation.
• This beta was validated primarily on Discord; other chat platforms may require agent-side output adaptation (image/message formatting differences).

Zx shorthand policy (project rule)

When user input is Zx, treat it as a strong render intent:

1. 1. Execute rendering directly by default (no preliminary Q&A).
2. Default path: CSS + minimal_ios_mobile + width=450.
3. Ask follow-up only under high uncertainty:

1. 4. If platform supports images, return the image directly (not link-only).

Data source priority for Zx:
1) current message image OCR
2) current message text-to-table
3) previous message image OCR
4) previous message text-to-table

Syntax sugar → canonical mapping

Pinned default baseline:

• - INLINECODE79
• INLINECODE80
• INLINECODE81
• INLINECODE82

page_spec rules

• - N: page N only
• INLINECODE85: inclusive range A..B
• INLINECODE86: from A to last page
• INLINECODE87: all pages
• if omitted: default preview pages INLINECODE88

Canonical payload example

CODEBLOCK0

Command examples

CODEBLOCK1

Validation checklist (minimum)

• - INLINECODE89
• CSS smoke output succeeds
• PIL smoke output succeeds
• INLINECODE90, --pin-reset, --both verified
• Golden tests pass when relevant

Release positioning

Current channel: beta.
ASCII remains beta/experimental for SkillHub release.

Support / Contact

• - GitHub Issues: https://github.com/con2000us/zenTable/issues
• Maintainer: @con2000us (Discord)
• Bug report checklist: