Lowkey Effort — AI skill for Claude Code

lowkey-effort 🧠⚡


An AI skill that auto-calibrates response depth based on task complexity — so you stop burning tokens on simple questions.

Works across **Claude, Claude Code, GitHub Copilot, OpenAI Codex, and Antigravity CLI**.

---

What it does

Most AI responses are either too long or too short. `lowkey-effort` fixes that by classifying every task into one of four effort tiers **before** responding, then applying the right format, depth, and length rules for that tier.

No more 500-word answers to "what's the capital of France?". No more shallow one-liners for complex system design questions.

---

The 4 Tiers

Tier	Target	When
⚡ Instant	< 50 tokens	Simple facts, yes/no, lookups
💬 Light	50–250 tokens	Short explanations, small snippets
📝 Standard	250–700 tokens	Coding tasks, guides, analysis
🔥 Deep	700–2000 tokens	Architecture, long-form, complex debug

The AI picks the tier silently — you won't even notice it's running.

---

Smart Overrides

The skill also listens for signals to upgrade or downgrade:

**Goes deeper if you say:** `"in detail"`, `"thorough"`, `"production-ready"`, `"comprehensive"`

**Stays light if you say:** `"quick"`, `"briefly"`, `"just"`, `"tldr"`, `"one-liner"`

**Never cuts corners on:** production code, security, health, or legal topics.

---

Installation

Claude

1. Download `lowkey-effort.skill`
2. In Claude, go to Settings → Skills
3. Click Install Skill and upload the .skill file
4. Done — activates automatically on every conversation

Claude Code

Drop the skill into your project:

your-project/
└── .claude/
    └── skills/
        └── lowkey-effort/
            └── SKILL.md

GitHub