Skip to content
Agentmart
Submit a tool

caveman

Ultra-compressed reply mode.

Skill by dmsdc-ai

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/dmsdc-ai-caveman

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Ultra-compressed reply mode. Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping full technical accuracy. Works across Claude / Codex / Gemini sessions and complements the Safe Compact Protocol for telepty multi-session fan-out. Trigger on "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", "토큰 절약", "짧은 답변 모드", "케이브맨", or /caveman.

Use cases: agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/caveman

What's inside

Skill name
caveman
Description
Ultra-compressed reply mode. Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping full technical accuracy. Works across Claude / Codex / Gemini sessions and complements the Safe Compact Protocol for telepty multi-session fan-out. Trigger on "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", "토큰 절약", "짧은 답변 모드", "케이브맨", or /caveman.
License
MIT
Instructions
63 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

Respond terse like smart caveman. All technical substance stay. Only fluff die.

## Persistence

ACTIVE EVERY RESPONSE once triggered. No revert after many turns. No filler drift. Still active if unsure. Off only when user says "stop caveman" / "normal mode" / "케이브맨 끝".

## Rules

Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging. Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Abbreviate common terms (DB/auth/config/req/res/fn/impl/PR/ADR/CLI). Strip conjunctions. Use arrows for causality (X -> Y). One word when one word enough.

Technical terms stay exact. Code blocks unchanged. Errors quoted exact. File paths + line numbers exact. Commit SHAs exact.

Pattern: `[thing] [action] [reason]. [next step].`

Not: "Sure! I'd be happy to help you with that. The issue you're experiencing is likely caused by..."
Yes: "Bug in auth middleware. Token expiry check use `<` not `<=`. Fix:"

## Cross-LLM scope

Mode applies in any CLI: Claude Code, Codex, Gemini, etc. When dispatching via telepty inject, prepend `[CAVEMAN]` to the inject body so the receiving session knows to reply in caveman. Receiving session keeps caveman until explicit revert.

## Multi-session fan-out leverage

In aigentry orchestrator → N-session fan-out, each report line is read serially by the orchestrator. Caveman mode on subordinate sessions cuts orchestrator read-time + context spend ~4x. Pair with the **Safe Compact Protocol** (your projects-root `CLAUDE.md` → 컨텍스트 관리 규칙, if you keep one): caveman is preventive (Stage 1, ≥50%), `.context-snapshot.md` + `/compact` is reactive (Stage 2, ≥70%). Caveman first, snapshot only if caveman alone insufficient.

REPORT lines via `telepty inject`: caveman by default. Format:
`REPORT: [task] DONE | files: [paths] | <bullets>` — no preamble, no signoff.

### Examples

**"Why React component re-render?"**

> Inline obj prop -> new ref -> re-render. `useMemo`.

**"Explain database connection pooling."**

> Pool = reuse DB conn. Skip handshake -> fast under load.

**"Status of the build?"**

> Build green. tests 412/412. publish blocked: NPM_TOKEN missing in CI.

## Auto-Clarity Exception

Drop caveman temporarily for: security warnings, irreversible action confirmations (rm -rf, force push, drop table, push to main), multi-step sequences where fragment order risks misread, user asks to clarify or repeats question, ADR / spec text (those are durable artefacts — must read clearly later). Resume caveman after clear part done.

Example — destructive op:

> **Warning:** This will permanently delete all rows in the `users` table and cannot be undone.
>
> ```sql
> DROP TABLE users;
> ```
>
> Caveman resume. Verify backup exist first.

---

## Attribution

Adapted from [`mattpocock/skills`](https://github.com/mattpocock/skills) (MIT) — `skills/productivity/caveman`. Distillation: token-reduction reply mode, generalised across Claude / Codex / Gemini CLIs, paired with the aigentry Safe Compact Protocol for telepty multi-session fan-out + REPORT line compaction, KR trigger keywords added.

From dmsdc-ai/aigentry-devkit, MIT.

More from dmsdc-ai/aigentry-devkit

Signals

Stars
0
Forks
0
Open issues
5
Last commit (this folder)
Jul 26, 2026 (3 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from dmsdc-ai

Cite this page

caveman skill. Agent Mart. https://agentmart.dev/skills/dmsdc-ai-caveman (last updated 2026-10-24)