Harriszed

Forgelite — Productivity skill for Claude Code

Productivity community

A lean, token-efficient build workflow for Claude Code: shape, plan, build, debug and verify with Opus where it matters and Sonnet where it’s cheap.

How to install Forgelite

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Harriszed/forgelite and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Forgelite does

A lean, token-efficient build workflow for Claude Code: shape, plan, build, debug and verify with Opus where it matters and Sonnet where it’s cheap. Built on Superpowers, at a fraction of the size and cost.

Alternatives in Productivity

Worktree Status Check — Verify the current worktree environment and show task details 23.4k ★
Performance Optimization Skill — Systematic performance analysis and optimization for RTK CLI tool, focusing on startup time (<10ms), memory us 11.9k ★
Ccplugins — Claude Code Plugins that actually save time 2.7k ★

README

$Forgelite: same workflow as Superpowers, a fraction of the tokens$

Forgelite

**A lean, token-efficient build workflow for Claude Code.** Shape an idea into a spec, plan it, build it test-first, debug by evidence, and prove it works before saying "done", while spending expensive model time only where it pays off.

Forgelite is built on the ideas of [**Superpowers**](https://github.com/obra/superpowers) by Jesse Vincent, which showed that a disciplined, skill-driven workflow makes Claude Code far more reliable. Forgelite was made with one goal: **keep that discipline, and improve speed, performance and token usage.** If you like what Superpowers does but want it to cost less and run faster, this is for you.

Forgelite is not affiliated with Superpowers or its author. See [Credits](#credits).

The short version

	Superpowers 7.0.0	Forgelite
Skills	15	6
Instruction lines	3,828	199
Text injected at every session start	~3,200 chars	320 chars
Independent reviews (spec, plan, final code)	yes	yes
Test-first building	yes	yes
Model chosen per subagent	mostly inherits your chat's model	yes (Haiku, Sonnet or Opus by job)
Opus for design and reviews, Sonnet for building	not built in	built in
Plain-language summaries aimed at non-coders	not a focus	yes