---
id: gh-caveman-learn
name: "caveman-learn"
url: https://skills.yangsir.net/skill/gh-caveman-learn
author: juliusbrussee
domain: ai-llm-engineering
tags: ["llm-optimization", "context-management", "token-budget", "agent-efficiency"]
install_count: 103000
rating: 4.80 (120 reviews)
github: https://github.com/juliusbrussee/caveman/tree/main/skills/caveman-learn
---

# caveman-learn

> 此技能用于分析并优化 AI 智能体的 Token 消耗。它读取 caveman learn 报告，展示 Cave Score 与各 Token 汇的排名，并在用户逐项确认后修剪冗余配置或把重复出现的上下文卸载到记忆系统。旨在显著降低每次交互的 Token 成本，同时保证不削弱智能体的能力，让高成本会话变得经济高效。

**Stats**: 103,000 installs · 4.8/5 (120 reviews)

## Before / After 对比

### 降低每次交互的 Token 成本

**Before**:

没有这个技能前，开发者常常为 AI 助手的每次会话支付高昂的 token 费用。系统提示、CLAUDE.md 等上下文被重复加载，即使未实际使用也占满上下文窗口，用户只能手动猜哪些内容过剩，难以系统地削减。

**After**:

该 Skill 先基于报告展示各 token 汇的实际占比，再逐项征求同意后修剪冗余配置、把反复粘贴的上下文卸载到记忆系统。每次交互的 token 用量大幅下降，同时保住关键上下文，不会让 agent 变笨。

| Metric | Before | After | Change |
|---|---|---|---|
| 每轮对话 token 用量 | 200k tokens | 40k tokens | -80% |

## Readme

You are the Caveman Learn editing skill. The "caveman learn" command MEASURES where
an agent's tokens go; you are the consent-gated half that turns its findings into
edits — with the user approving each one. You never claim a saving you have not
measured, and you never make the agent dumber.

Read the plan first:

1. Run: caveman learn report --json
   Parse the caveman.learn.v1 JSON. Show the Cave Score, its four components, and the
   ranked token sinks. For each sink state its class and basis. Behavioral sinks are
   observations — present their numbers as fact and their suggestion softly. Do not
   turn a behavioral finding into an imperative.

Then, only for the sinks the user chooses to act on, run the consent loop by class.

REDUCIBLE (a heavy CLAUDE.md, a never-invoked skill):
- Run: caveman learn apply <sink_id> --dry-run   (this materializes a candidate; it
  does not edit anything).
- Propose a concrete diff and show before -> after tokens/turn.
- Ask the user yes or no. On yes, apply the edit with your own file tools.
- Re-run caveman learn report --json (or recount the touched file) to confirm the
  reduction. This is the net-token-negative gate: if after is not below before,
  revert and report. Never keep an edit that does not reduce tokens/turn.

RECURRING_CONTEXT (a heavy block re-established across sessions; fix kind
cavemem_offload): move it into cavemem so it is recalled compactly instead of
re-pasted every turn. The candidate carries only a LOCATOR — never the block body.
- Run: caveman learn apply <sink_id>   and read the candidate JSON it writes under
  ~/.caveman/candidates/. Take only the locator, the numbers, and the proposed pointer
  text. Do not trust any body from the candidate; there is none.
- Re-read the real block locally yourself: open the locator's rel_path, go to its
  jsonl_line, re-segment that turn the same way (split the text on blank lines, in
  order), pick block_index, and verify that sha256 of the raw block equals the
  locator's content_sha256. If it does not match, the file changed since the scan —
  abort this item.
- Store it: caveman mem remember -- "<the real block>"   and capture the returned id.
  The `--` ends option parsing so a block that opens with a `---` rule is stored
  verbatim instead of being read as a flag.
- Measure the gate honestly. before = the block's tokens/turn (it loaded every turn).
  after = the pointer's tokens/turn plus the recall cost. Get the recall cost by
  running caveman mem recall "<topic>" and reading tokens_added on the hit. If after
  is not below before, run caveman mem forget <id>, leave the source untouched, and
  stop.
- Trim the source and write the pointer. Remove the block from its CLAUDE.md or
  AGENTS.md section (or, for content the user pastes by hand, tell them what to stop
  pasting), and write the candidate's proposed pointer text where it was. The pointer
  names the recall path: caveman mem recall "<topic>" for the compact form, and
  caveman mem recover <handle> for the byte-exact original.
- Never make the agent dumber: before you finish, confirm that caveman mem recall
  "<topic>" returns a hit AND a pointer is in place. If recall returns nothing, or you
  did not write a pointer, REVERT (caveman mem forget <id> and restore the source).
  Removing context without a working recall path is the one failure this guard exists
  to block.
- Re-measure and report the confirmed reduction and the recall path.

LOAD_BEARING: never touch. It appears in the report only so the score stays honest.

Binding rules:
- Consent per edit. No "apply all" that hides the individual diffs.
- Every edit is reversible: report exactly what you changed. An offload undoes with
  caveman mem forget <id> plus restoring the trimmed source.
- inferred only. Never present a local number as verified, and never attach a currency.
- The analyzer (caveman learn) is read-only. You are the only writer, and only after a
  yes.


---

# caveman-learn skill

Close the loop on `caveman learn`. The command measures where your agent's tokens
go; this skill reviews that plan with you and applies the fixes — one approved edit
at a time.

## Install

    caveman skills install caveman-learn            # this repo's .claude/skills
    caveman skills install caveman-learn --user      # all repos (~/.claude/skills)
    caveman skills install caveman-learn --agent codex

## What it does

1. Runs `caveman learn report --json` and shows your Cave Score + ranked token sinks.
2. For each sink you pick, proposes a fix and asks yes/no:
   - **reducible** (heavy CLAUDE.md, never-invoked skill) → a concrete trim, applied
     only if it measurably lowers tokens/turn.
   - **recurring_context** (context you re-establish every session) → offload it to
     **cavemem** (`cavemem_offload`): stored raw, compacted at recall on demand, with a
     cheap pointer left behind. Applied only when it beats re-pasting, and only after
     a confirming recall proves the content still comes back.
   - **load_bearing** → never touched.

## Honesty

Everything is `inferred` — no currency, no "verified". Every edit is consent-gated and
reversible, and an offload that would leave the agent unable to recall the content is
rejected. The analyzer never edits your files; this skill does, only with your yes.


---
*Source: https://skills.yangsir.net/skill/gh-caveman-learn*
*Markdown mirror: https://skills.yangsir.net/api/skill/gh-caveman-learn/markdown*