brenbuilds1/cutweight/cutweight-weigh

Weigh-In

The scale for your agent. cutweight-diet finds what to cut; cutweight-weigh tells you what you weigh, before the cut and after it. Step on the scale first, cut, step on it again.

Method

1. Measure the always-on bill exactly as cutweight-diet does: installed skills

(frontmatter name plus description each), memory files loaded at session start (CLAUDE.md, AGENTS.md, local variants, imports), MCP tool schemas (eager-loaded only; a server you cannot enumerate is unknown, never zero).

1. Rough tokens for English markdown: characters / 3 on Claude models with

the Opus 4.7 tokenizer (Opus 4.7+, Sonnet 5, Fable 5), characters / 3.5 to 4 on older Claude and GPT-style tokenizers. Label estimates.

1. Read the last card from .cutweight/weigh-ins.md if it exists. Compute

the delta.

1. Print the card. Append it to the log. Never edit old cards.

Weight Classes

Cutweight's own scale, by always-on tokens per message:

The Card

cutweight weigh-in: 2026-07-10
always-on: 7,412 tokens on every message
  skills: 3,509 (27 installed)
  memory files: 3,903
  mcp schemas: ~0 (deferred loading)
class: middleweight
last weigh-in: 9,180 (-19%, was heavyweight)
next cut: the 25-skill plugin billing 2,462 for 4 skills used

Keep the card under 12 lines. It is made to be posted.

Hard Rules

• The log is append-only. A rewritten history is not a weigh-in.
• Measured or labeled-estimate numbers only.
• The card always names the single next cut, or says "at weight" if lean.
• Never name an agent-maintained memory file (auto-memory index, session

memory) as the next cut. Bill it on the card, flag it to its owner. Its lines are the agent's retrieval keys; hand-trimming them breaks recall.

• A card meant for posting shows counts and categories, never private tool,

server, or plugin names. Generalize anything internal before it leaves the machine.

• Keep .cutweight/ out of version control; the log stays on the machine.
• Weigh-in changes nothing but its own log. The human does the cutting.