A set of rules for keeping work within a limited token budget, where tokens are the text units used by an AI assistant. It emphasizes planning, focused reading, careful questions, and verification.

Use it for large multi-step tasks, bulk file or data reviews, subagent work, or whenever you need to control usage and cost.

It reduces wasted assistant output and repeated work while preserving checks that prevent errors.

lean — Skill for Claude Code and Codex by nataliacorrea03

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 nataliacorrea03/claude-code-skills --skill lean

Clone the repo

git clone --depth 1 https://github.com/nataliacorrea03/claude-code-skills

Made for: Claude Code, Codex.

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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/nataliacorrea03/claude-code-skills/lean/github.svg)](https://agentmods.dev/skills/nataliacorrea03/claude-code-skills/lean)

Your own site

<a href="https://agentmods.dev/skills/nataliacorrea03/claude-code-skills/lean"><img src="https://agentmods.dev/badge/skills/nataliacorrea03/claude-code-skills/lean/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/nataliacorrea03/claude-code-skills/lean"><img src="https://agentmods.dev/badge/skills/nataliacorrea03/claude-code-skills/lean.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 75 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 1,168 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.

Model	Per session	Once invoked
Fable 5.1	$0.00075	$0.01168
Opus 5.5	$0.00030	$0.00467
Sonnet 5.5	$0.00015	$0.00234
Haiku 4.5	$0.00007	$0.00117

Measured 2mo ago against content hash 3ee9e8e07ccf, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

lean 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 2mo ago.

The scan reads SKILL.md. This mod also ships 1 executable file (scan-sessions.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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/lean/SKILL.md · 74 lines

How it starts

The opening of the file, as written. The whole thing — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Lean Mode

Core principle: quality stays, waste goes. Token efficiency means eliminating redundant work, never cutting corners. A wrong guess, a skipped verification, or a sloppy edit costs more tokens to fix than it saved.

First action on invocation: read LEARNINGS.md in this skill's folder and apply every entry. It is short by design.

The Protocol (execute in this order)

1. Plan first

Write a numbered plan (7 steps max) before any multi-step work. Execute chronologically. Re-plan only when blocked, not when curious.

2. Ask, never guess

If anything is ambiguous, ask BEFORE starting: one batched AskUserQuestion with multiple-choice options covering ALL open questions. One round trip. A wrong assumption doubles the job.

3. Intake discipline (input tokens)

Grep/glob before Read. Read line ranges, never whole large files.
Never re-read a file you just edited or already have in context.
Big command output: pipe through head/tail/grep/jq, or redirect to a scratch file and grep that.
MCP pulls (Notion, Gmail, Drive, project trackers): fetch the specific page/thread/record. Never dump whole databases or unfiltered search results.
ToolSearch: load every tool you'll need in ONE call.
Browser work: prefer text extraction over screenshots. Images cost far more tokens than text.
Notes and memory files are leads: use them to skip broad discovery, then verify only the specific fact live.

4. Delegation policy

NEVER spawn a subagent on your most expensive model tier. Set an explicit cheaper model on every Agent call: a mid-tier model by default, the cheapest tier for mechanical work (bulk reads, copy transforms, scraping, formatting).
Every subagent gets a complete plan: exact file paths, exact output format, no exploration budget.
Don't delegate what one grep answers. Do delegate bulk surface work so the expensive main model stays the orchestrator.

5. Output discipline (output tokens cost ~5x input)

Edit with diffs. Never rewrite a whole file for a partial change.
Never echo file contents or code back into chat. Reference paths as file:line.
Final summary: 5 sentences or fewer unless the task genuinely needs more. No restating the plan or the diff.
Build only what was asked. No speculative extras, no unrequested refactors.

Read the full file on GitHub · 74 lines

Files

What ships with it

3 files 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.

2mo ago First seen · 74 lines · 75 tokens per session scan A 3ee9e8e07ccf

Subscribe to this mod's changes

lean is a skill published in the GitHub repository nataliacorrea03/claude-code-skills (4 stars, last pushed 3mo ago), licensed MIT. It adds 75 tokens to every session and 1,168 once invoked, about $0.0003 per session on Opus 5.5. 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.

Claude Code LLM Productivity agent-skills ai-agents claude developer-tools

lean next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
leannataliacorrea03/claude-code-skills	Skill	4	3mo ago	MIT	A	75	Claude Code, Codex
apple-remindersNousResearch/hermes-agent	Skill	262k	today	MIT	A	17	Claude Code, Codex
apple-notesNousResearch/hermes-agent	Skill	262k	today	MIT	A	17	Claude Code, Codex
boxNousResearch/hermes-agent	Skill	262k	today	MIT	A	14	Claude Code, Codex
obsidianNousResearch/hermes-agent	Skill	262k	today	MIT	A	19	Claude Code, Codex
channels-setupCopilotKit/CopilotKit	Skill	39k	today	MIT	A	141	Claude Code
mantis-configuregoogle/mantis	Skill	2.7k	3d ago	Apache-2.0	A	77	Claude Code, Codex