Skip to content
Back to skills

Tactics Stealth

ASecurity

Use when a real-time squad or multi-actor stealth game must compose perception, information, planning, movement and ranged/melee actions without hardcoding one specific cone model or command cadence.

  • 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 tactics-stealth --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Tactics Stealth?

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

Security grade badge for Tactics Stealth
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jammyfu-tactics-stealth/badge)](https://www.skillsdirectory.com/skills/jammyfu-tactics-stealth)

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: tactics-stealth
description: Use when a real-time squad or multi-actor stealth game must compose perception, information, planning, movement and ranged/melee actions without hardcoding one specific cone model or command cadence.
---

# Tactics stealth

Ask which project modes already exist: perception model, information presentation, planning/queue mode, movement/stance model, ranged/melee action owners, and alert-sharing policy.

## Contract

Publish:
- stable `stealth_scenario_id` and revision
- actor/squad identities and authored role capabilities
- perception policy reference from `enemy-perception`
- information/presentation policy reference from `stealth-info`
- planning/commit policy from `plan-queue` when commands are queued
- action owners for movement, melee and ranged resolution
- alert/noise propagation policy and stable event identity
- fail/recovery/reset policy

## Ownership

`tactics-stealth` owns composition of a multi-actor stealth scenario. It does not redefine sight/hearing truth, shot resolution, queue timing, locomotion, HUD cones or enemy target selection.

## Rules

1. Inner/outer cones, continuous suspicion, light/noise fields or other perception models are project data; dual-cone stealth is not universal.
2. Crawl/stand visibility, stance speed and cover behavior come from the owning perception/locomotion policies rather than genre assumptions.
3. A noise or alert event affects only recipients selected by the authored propagation policy; one loud action does not universally alert the whole map.
4. Queued commands follow `plan-queue` commit ordering. They are not required to start on the same logical tick.
5. Melee, ranged, enemy-fire and targeting keep their own stable action/event identities and clocks.
6. Tactical UI may visualize authorized information but cannot grant NPCs knowledge or modify perception truth.

## Acceptance

Given the same scenario revision, actor state, perception/alert inputs and queued command set, the same scenario-level transitions and delegated action requests are produced. Changing tactical UI presentation or render cadence does not change who perceived what, which commands committed, or the authoritative action outcomes.

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.