claude

lean

Cuts the tokens and time of long sessions that fan out subagents and workflows: agents at a fixed model and effort, and a report of what a session costs, by tier, model and cause.

0 on GitHubby Aleksandr Zadykianv7.1.0MITupdated yesterdayproductivity
/

Plugin lean from marketplace zadykian in zadykian/toolkit. Commands read from the repo's manifest — today.

What it installs

Skills (20)

  • atomic-commits

    Use when implementing code changes in a git repo - deliver small, self-contained Conventional Commits, and split any change over 500 diff lines into a stack of up to 5 independently mergeable PRs.

  • banned-terms

    Keep vague names and fillers, such as helper, utils, misc and simply, out of new code and prose. Use when naming a file, type, function or variable, and when writing a comment, a doc, a commit message, or a pull request's title and body. - "**/*.md" - "**/*.go" - "**/*.py" - "**/*.sh" - "**/*.js" -

  • ci

    Write a repository's GitHub Actions workflows: short jobs that run on every pull request, long jobs that wait for a ready stack on the default branch, and what every job sets. Use when writing or editing a workflow, or when CI's runners queue.

  • codebase-design

    Give the shared vocabulary for deep modules: a lot of behaviour behind a small interface, at a clean seam. Use when designing or improving a module's interface, finding where to deepen, placing a seam, or making code easier to test, or when another skill needs the terms.

  • context

    Keep a long session's context small: how to read, how to bound output, when to write a checkpoint and compact, and how long a report runs. Use in a session that reads many files, runs many commands, or has run for a long time.

  • cost

    Report what Claude Code sessions cost, from their transcripts: by tier, model and context size, cache writes by idle gap, the largest tool results, foreground polls, and each workflow's real tokens. Run by hand, to measure a baseline and the change after.

  • delegate

    Decide whether to hand work to a subagent, which agent takes it, how to write its brief, and how to read what it returns. Use before each spawn of an agent, and before a batch of them.

  • domain-modeling

    Build and sharpen a project's domain model: challenge its terms, keep its glossary, and record its hard decisions. Use when discussing a codebase's terminology, or writing or editing a glossary or a decision record.

  • fan-out

    Size review by risk, tier a workflow's agents, and keep its results small. Use when writing a Workflow script, or when planning to review, verify or fix a change with several agents.

  • grill-me

    Interview the user, round by round, until a plan, decision or idea is understood by both. Use when the user asks to be grilled, or wants to stress-test their thinking.

  • grill-with-docs

    Interview the user to sharpen a plan or design, and keep the glossary and the decision records as the talk goes. Use when the user asks to be grilled and to have the terms and decisions written down.

  • improve-codebase-architecture

    Scan a codebase for deepening opportunities, write them up as a Markdown report, then grill through the one the user picks. Use when the user asks to improve the architecture, or to find shallow modules to deepen.

  • issue

    Plan a change, then file it as a GitHub issue in a compact house style once the maintainer approves the plan. Use when asked to file, open or write up an issue, or to plan a change as one.

  • land

    Land a stack of approved pull requests on main, bottom first, by GitHub's merge commits, and tag a release if asked.

  • repo-settings

    Check a GitHub repository's merge settings, rulesets and Actions permissions against those that stacked pull requests land by, report each difference, and apply them once approved.

  • split-file

    Split a source file over the line cap into files by responsibility, in its package or module and with its API unchanged. Use when a size check reports a file over its cap.

  • tdd

    Build a feature or fix a bug test-first, in red-green slices at agreed seams, with tests that state behaviour and survive a refactor. Use when asked to work test-first, to run red-green-refactor, or to add integration-style tests.

  • trim

    Trim a repository's code comments or Markdown docs to the writing policy and its Vale rules, keeping every fact. Use on files that fail Vale or the policy's caps, or that restate the code or the help.

  • wait

    Wait on work without holding the turn or the cache: run it in the background, watch for a change of state, and bound a foreground wait. Use when starting a build, a test run, a CI run or a deploy, when about to poll, or when scheduling a wake-up.

  • writing

    The writing policy for code comments and docs: a comment gives a purpose or a reason, each fact lives in one place, and short sentences pass Vale's rules. Use when writing or editing comments, Markdown or other prose in a repository. - "**/*.md" - "**/*.go" - "**/*.py" - "**/*.sh" - "**/*.js" - "**/

View source on GitHub

More from zadykian/toolkit

All 5 plugins →

design

claude

Skills that sharpen a design before code: an interview, a glossary and decision records, a deep-module vocabulary, test-first slices and an architecture scan.

20 skills

0

docs

claude

A writing policy for comments and docs that loads as Claude edits, and a skill that trims files to it, keeping every fact.

20 skills

0

flow

claude

From issue to main: plan a change as an issue, commit it in small Conventional Commits and stacked pull requests, check it in CI a stack at a time, and land the stack by GitHub's merge.

20 skills

0

lint

claude

Hooks that run the repository's own lint gates on each file Claude edits and on the branch when it stops, a skill that splits a file over the line cap, and one that keeps vague names and fillers out of new code and prose.

20 skills

0

Similar plugins

superpowers

claudecodex

Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques

15 skills · hooks

obra/superpowers 316,836

caveman

claude

Talk like caveman. Cut filler. Keep technical details explicit.

22 skills · 1 command · 3 agents · hooks

JuliusBrussee/caveman 119,462

claude-mem

claudecodex

Persistent memory for Claude Code: the next session starts where the last one left off.

29 skills · 1 MCP server · hooks

thedotmack/claude-mem 116,618

Taste Skill

claudecodex

tasteskill 2.5.1: professional company websites at the figma.com and x.ai level, with one section per screen, real photographs, real product UI, effects kept for tech and gaming pages, and a real-browser checker and finish gate.

1 skill

Leonxlnx/taste-skill 114,391

career-ops

claudecodex

Open-source AI job search that runs locally in your AI coding CLI: scan job portals, evaluate listings into a structured A-H report with a 1-5 score, tailor your CV, track applications

5 skills · hooks

career-ops-hq/career-ops 82,253

mem0

claudecodex

Cross-session memory and token savings for coding agents.

6 skills · 1 agent · 1 MCP server · hooks

mem0ai/mem0 68,893

Maintain lean?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex plugins5 plugins