skill-diet

skill-diet is a skill for Claude Code, Codex from brenbuilds1/cutweight. It costs 105 tokens per session (1,134 once invoked), scanned A, original, MIT.

A method for measuring and reducing the always-loaded context used by an agent’s skills, memory files, and tool definitions.

In plain words
What is it for?
Use it to audit installed skills, AGENTS.md or CLAUDE.md files, and MCP tool schemas, then decide what to trim or move out of the default setup.
Why use it?
It shows which setup parts consume context even when unused, helping keep the agent focused as the configuration grows.

Skill for Claude CodeCodex

Written for Claude Code and Codex: allowed-tools in frontmatter, but also agents/openai.yaml present. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions AGENTS.md.

Part of the cutweight plugin — 5 skills shipped together

Good fit Use it to audit installed skills, AGENTS.md or CLAUDE.md files, and MCP tool schemas, then decide what to trim or move out of the default setup.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/brenbuilds1/cutweight/skill-diet
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add brenbuilds1/cutweight --skill skill-diet
Clone the repo
git clone --depth 1 https://github.com/brenbuilds1/cutweight

Made for: Claude Code, Codex.

Or install cutweight, the plugin that ships this one along with the rest of its 5 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for skill-diet

README.md
[![agentmods](https://agentmods.dev/badge/skills/brenbuilds1/cutweight/skill-diet/github.svg)](https://agentmods.dev/skills/brenbuilds1/cutweight/skill-diet)
Your own site
<a href="https://agentmods.dev/skills/brenbuilds1/cutweight/skill-diet"><img src="https://agentmods.dev/badge/skills/brenbuilds1/cutweight/skill-diet/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for skill-diet

Your own site · 80×15
<a href="https://agentmods.dev/skills/brenbuilds1/cutweight/skill-diet"><img src="https://agentmods.dev/badge/skills/brenbuilds1/cutweight/skill-diet.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 105 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,134 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00105 $0.01134
Opus 5 $0.00053 $0.00567
Sonnet 5 $0.00021 $0.00227
Haiku 4.5 $0.00011 $0.00113

Measured 8d ago against content hash bf8b1f2bdeb1, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

skill-diet scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/skill-diet/SKILL.md · 95 lines

How it starts

The opening of the file, as written. The whole thing — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Skill Diet

Skill descriptions, memory files, MCP tool schemas: your agent reads all of it on every message before any work happens. Caching makes that cheap in dollars. It still fills the context window, and full windows measurably degrade output. This skill reads the whole bill.

When To Use

After installing a skill pack or connecting an MCP server. When the agent picks the wrong skill. When the bill grows without more work done.

Method

Rough tokens for English markdown: characters / 3 on Claude models carrying the Opus 4.7 tokenizer (Opus 4.7+, Sonnet 5, Fable 5), characters / 3.5 to 4 on older Claude (Sonnet 4.6, Haiku 4.5) and on GPT-style tokenizers. The newer Claude tokenizer counts roughly 30% higher, varying by content, so never reuse counts across models. When exact matters, Anthropic's count-tokens endpoint is free and per-model; for GPT, tiktoken runs locally. Label every number measured or estimated.

  1. Skills. Find every installed skill: ~/.claude/skills, .claude/skills, ~/.codex/skills, .agents/skills, plugin directories. Record two numbers each: always-on (frontmatter name plus description, injected into every request) and on-trigger (body plus every file it tells the agent to read).
  2. Memory files. Find everything loaded at session start: CLAUDE.md and AGENTS.md at user level, project level, subdirectories, local variants, plus files they import. All always-on. Measure each. Leave style linting to linters; the job here is the bill. Files a human wrote (CLAUDE.md, AGENTS.md) are fair to trim. Files an agent maintains for itself (auto-memory indexes, session memories) get billed, NEVER hand-trimmed: their lines are retrieval keys, and cutting them breaks the agent's recall in ways that surface weeks later. If one must shrink, the owning agent compacts it in-context.
  3. MCP tool schemas. Usually the biggest line item: each connected server injects its tool catalog into every request, commonly 500-1,400 tokens per tool. List servers from .mcp.json and client settings, count each server's tools, estimate schema cost. If the client defers tool loading (tool search, lazy loading), bill only what loads eagerly and say so. A server you cannot enumerate is unknown, never zero.
  4. Overlap and dead weight (skills). Name pairs claiming the same trigger phrases. If session history exists, list skills that never fired; otherwise ask the user and mark the rest unknown, not unused.
  5. Read the fattest items. Rules earn their tokens. Backstory does not.

Read the full file on GitHub · 95 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago First seen · 95 lines · 105 tokens per session scan A bf8b1f2bdeb1

Subscribe to this mod's changes

skill-diet is a skill published in the GitHub repository brenbuilds1/cutweight (2 stars, last pushed 1mo ago), licensed MIT. It adds 105 tokens to every session and 1,134 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

sdd

Execute the Liatrio Spec-Driven Development (SDD) workflow when explicitly invoked by the user. NOTE: this skill is NOT intended to be dynamically loaded or automatically triggered; it should only ever be explicitly called by the user.

liatrio-labs/spec-driven-workflow · 50 tokens

bcc-plan-spar

BCC align+lock+review PLAN.md for one slice (no product code). Slash: /bcc-plan-spar · chat: bcc:plan-spar · "lock PLAN" · spar the plan. Args: rounds=N (auto-review cap), review=self|subagent|cli|auto|off. Grill until clear enough (no default Q&A quota). Hand off to bcc-clean-cut after human APPROVE.

bo-cao/breaking-coding-chaos · 91 tokens

bcc-throughline

BCC global progress cockpit (plans.md/progress.md/findings.md). Slash: /bcc-throughline · chat: bcc:throughline · "where are we" · reprioritize · resume after /clear. Not for coding or full PLAN grill.

bo-cao/breaking-coding-chaos · 56 tokens

bcc-breaking-coding-chaos

BCC main skill: dual-loop coding workflow (throughline → plan-spar → clean-cut) or quick status+next. Slash: /bcc-breaking-coding-chaos · chat: bcc:breaking-coding-chaos · "run BCC" · "BCC status" · "what next BCC". Needs a real idea (1:1 implement). Args: goal text, or status. May pass plan-spar review budget as…

bo-cao/breaking-coding-chaos · 101 tokens

repo-layout

Register, resolve, and validate where a project's path-slots actually live via .layout-slots.yaml markers, so relocating a slot needs zero SKILL.md edits. Do NOT use for single-file lookups.

linkpranay-ai/context-engineering-protocol · 45 tokens

compiling-project-guidelines

Compile scattered guideline sources into one scope-aware COMPILED-GUIDELINES.md for other skills and ult-context-generate's Constraints layer. Do NOT use to enforce rules at runtime.

linkpranay-ai/context-engineering-protocol · 43 tokens