agentleFS
Sign inStart free
Co4anCo4an/claude-auto/skills/lean/SKILL.md

lean

Skill0 starsChanged 6 months agoMITView on GitHub

Switches the current session into token-economy mode for the next response/task. Forces Haiku-first, forbids large file reads, restricts to single-agent delegation, no synthesis stage. Use when budget is tight or when the task doesn't deserve premium tokens.

9 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use ship.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AWSSkill

    ship

    Finalize dev work on an issue branch — update docs, push, and open a PR.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No web research beyond 3 searches: If 3 searches don't answer it, write "not findable in 3 searches under /lean — try /auto for deeper …

    line 21
  2. Very specific:

    Read the task from $ARGUMENTS.

    line 26
  3. Very specific:

    Refuse to escalate to Opus: If task truly needs Opus, abort and recommend /auto opus.

    line 49
  4. Very specific:

    No deep research orchestration: If the user asks for "deep research" under /lean, abort and recommend /auto.

    line 51

“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

LEAN — Token-economy mode

You are operating in token-economy mode for this task. The goal is to deliver a 70-80% quality answer at 20-30% of the normal token cost.

Rules for this session

  1. Default model = Haiku for any sub-agent you spawn (unless the task strictly cannot be done on Haiku — then escalate to Sonnet, never Opus).
  2. No multi-agent pipeline. Maximum ONE sub-agent. If the task needs more, tell the user "this task is too big for /lean — recommend /auto instead" and stop.
  3. No large file reads. Refuse to Read any file >500 lines. If the user asks, use Grep first to find the relevant section, then Read with offset + limit.
  4. No synthesis stage. Single agent output goes directly to user, no separate synthesist.
  5. No fact-checking pass. Trust the worker; flag confidence inline ("single-source, unverified").
  6. No web research beyond 3 searches. If 3 searches don't answer it, write "not findable in 3 searches under /lean — try /auto for deeper investigation."
  7. Output cap: 200 lines. Tighter than normal.

Process

  1. Read the task from $ARGUMENTS.
  2. Decide: is this answerable with a single Haiku agent in <5 searches?
    • Yes → pick the single best agent, spawn on Haiku, return its result with confidence labels.
    • No → respond: "Эта задача требует более серьёзного pipeline. Запускать /auto?"

Output

Direct, no preamble. Confidence labels on claims. If anything is uncertain, say so.

[Direct answer]

[Inline sources where available]

**Confidence:** [high/medium/low]
**Limits of this lean answer:** [what wasn't checked]

---
Если нужна глубже — `/auto <task>`.

Hard rules

  1. Refuse to escalate to Opus. If task truly needs Opus, abort and recommend /auto opus.
  2. No background subagents. Lean mode is foreground only.
  3. No deep research orchestration. If the user asks for "deep research" under /lean, abort and recommend /auto.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free