Skip to content
Back to skills

Lock On Target

ASecurity

Use when a character needs soft or hard target locking with stable target/session identity, deterministic candidate selection, retention/drop rules and explicit consumers for facing, attacks, HUD and camera.

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

Works with

  • cli

Security analysis

A100/100

Scanned October 23, 2026

npx -y skills add jammyfu/open-game-skills --skill lock-on-target --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Lock On Target?

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

Security grade badge for Lock On Target
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jammyfu-lock-on-target/badge)](https://www.skillsdirectory.com/skills/jammyfu-lock-on-target)

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: lock-on-target
description: Use when a character needs soft or hard target locking with stable target/session identity, deterministic candidate selection, retention/drop rules and explicit consumers for facing, attacks, HUD and camera.
---

# Lock-on target

Ask: `soft-lock | hard-lock | none`.

## Ownership

`lock-on-target` owns selected target identity and lock lifecycle. Target eligibility comes from authoritative target/perception state. Camera, facing, attack aim and HUD may consume the selected target according to their own modes; they are not required to share identical behavior in every project.

## Contract

Publish:
- stable `lock_session_id`
- selected `target_id`
- candidate-set / targeting revision
- acquire/retain/switch/drop reason
- authored ranking, tie-break and hysteresis policy
- consumer policy for camera/facing/attack/HUD

Cycle input is a semantic action from `input-design`; mouse/stick bindings are not hardcoded here.

## Runtime rules

1. Candidate ordering is deterministic for equal scores and never depends on render iteration order.
2. Visibility, range, state or other eligibility changes are evaluated from the targeting owner, not from the lock pip.
3. A stale target revision cannot restore a dropped target.
4. Switching produces one new selected target per committed request; duplicate cycle requests are idempotent when they share request identity.
5. Camera assistance still obeys camera ownership/collision; lock selection never bypasses `camera-anti-clip`.

## Acceptance

Given the same candidate revision, actor state and switch request, the same target is selected or no target is returned. HUD/camera presentation can change independently without changing target identity, and a removed/ineligible target cannot persist as a ghost lock.

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.