Skip to content
Back to skills

Interact Prompt

ASecurity

Use when the HUD must present currently legal world interactions, semantic actions and current bindings without owning interaction truth, target priority, localization identity or gameplay execution.

  • 11 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 23, 2026
ai-agents

Security analysis

A100/100

Scanned October 23, 2026

npx -y skills add jammyfu/open-game-skills --skill interact-prompt --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Interact Prompt?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Interact Prompt
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jammyfu-interact-prompt/badge)](https://www.skillsdirectory.com/skills/jammyfu-interact-prompt)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: interact-prompt
description: Use when the HUD must present currently legal world interactions, semantic actions and current bindings without owning interaction truth, target priority, localization identity or gameplay execution.
---

# Interact prompt

Ask: `context-one | listed-verbs | no-prompt`.

## Ownership

The interaction/gameplay resolver supplies eligible actions, target IDs and authored priority. `input-design` supplies semantic action bindings. `game-localization` supplies stable string IDs/text. This skill only chooses and presents the authorized prompt set.

## Contract

For every prompt resolve:
- stable `target_id`
- stable semantic `action_id`
- source resolver revision
- current binding glyph/text
- stable localization string ID
- visibility/occlusion eligibility

Do not hardcode a universal interaction ordering. If multiple interactions are legal, use the project's authored priority/tie-break.

## Runtime rules

1. A stale, removed or occluded target invalidates its prompt before activation.
2. Prompt updates are idempotent by target/action/revision.
3. Focus/target churn uses project hysteresis or tie-break data; render frame order is not authority.
4. Changing language or physical binding updates presentation without changing `action_id`.
5. Executing an action belongs to the gameplay owner; showing a prompt never commits it.

## Acceptance

Before activation, the UI can identify the exact target/action/revision it is presenting. If the target disappears or a newer resolver revision wins, the stale prompt cannot fire, and remapping/localization changes never alter the underlying gameplay verb.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in ai-agents →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.