alebgl77

Harnessmeter — Development skill for Claude Code

Development community

A profiler for your agentic harness.

How to install Harnessmeter

This entry records only its repository, not the path inside it, so there is no exact command to give. Open alebgl77/harnessmeter and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Harnessmeter does

A profiler for your agentic harness. Price every line of CLAUDE.md, every skill, subagent and MCP tool schema — then find the ones that don't earn their place. Zero API keys, zero network. Claude Code, Codex, Antigravity.

Alternatives in Development

  • Sanitize — Detect and redact PII from text files — 15 categories (SSNs, credit cards, API keys, etc.), zero dependencies 3.1k ★
  • Dao Code — Open-source TypeScript terminal coding agent for DeepSeek-V4 — builds on DeepSeek's strong price-performance a 1.3k ★
  • Backpass — You don't write AGENTS.md 1.1k ★

README

harnessmeter

npm ci status: early node license: MIT zero dependencies zero api keys no network


You have 4,000 lines of `CLAUDE.md`, twelve subagents, eight MCP servers, and **zero tests**. Someone edits one line and the behaviour of your whole team shifts, silently.

We have profilers for CPU, memory, SQL, and JS bundles. We have **none for context** — the scarcest resource in agentic engineering. Every token of your harness is loaded on *every turn*, of *every session*, of *every developer*. Nobody can tell you which lines pay for themselves.

`harnessmeter` prices them.

The frame

**Your context window is a commons with no rent.** Every instruction occupies space for free, forever, regardless of what it produces. Two things follow, and you've felt both:

  • CLAUDE.md only ever grows.
  • At 4,000 lines, nobody dares delete anything, because nobody can prove what's load-bearing.

So we give every block a **lease**: a measured price, a measured yield, a