Skip to main content

TableauQuest for Codex

TableauQuest is a choice-driven simulation skill that teaches Tableau judgment through realistic technical and organizational consequences.

Created by Adam Mico.

Features​

  • Single scenarios, campaigns, and Field Quests.
  • Blitz through Deep session lengths.
  • Consequence-driven choices without obvious correct answers.
  • Tableau-realistic constraints spanning performance, governance, metrics, reliability, security, UX, and organizational trust.
  • Optional, consented read-only grounding through a connected Tableau MCP server.
  • Text-only pixel-art endings and a judgment-based Sparkle evaluation.

Requirements​

  • Codex.
  • Optional Tableau MCP connection for live-grounded Field Quests.

The skill runs fully with fictional scenarios when no Tableau connection is available.

Example prompts​

Use $tableau-quest to start a quick Tableau adventure.
Start TableauQuest in noir mode. I am an advanced developer and want a metric-conflict scenario.
Use $tableau-quest for a Field Quest based on my Tableau project. Anonymize all people and keep the scan read-only.

Package contents​

  • SKILL.md — Codex entrypoint and gameplay contract.
  • agents/openai.yaml — discovery metadata and optional Tableau MCP dependency.
  • skill-expertise/scenario-design.md — scenario archetypes, realism, choices, and calibration.
  • skill-expertise/tableau-grounding.md — consent, MCP discovery, evidence semantics, and privacy.
  • skill-expertise/state-and-endings.md — state, campaigns, interruptions, Sparkle, and endings.
  • skill-expertise/validate_state.py — validator for exported/restored quest state.
  • skill-expertise/test_validate_state.py — validator unit tests.

Safety and privacy​

Field Quest uses only the minimum read-only metadata needed. It does not place real owners into fictional negative roles, expose raw responses, or modify Tableau content. It distinguishes observed site facts from fictional story details.

State​

Quest state remains in the current conversation by default. The skill does not automatically write memory or files. Users may explicitly export structured state and validate it with:

python skill-expertise/validate_state.py state.json

Validation status​

Local package and unit tests can verify structure and state invariants. A live forward test requires a separate Codex session connected to Tableau MCP.

License​

No license is asserted by this conversion. Add one only if you have the right to do so.