Why paleo?

• Tokens cost money and latency. Every trimmed token means faster, cheaper inference.
• One-size-fits-all prompting fails. Sometimes you want terse output, sometimes a hard budget, sometimes just no fluff. paleo gives each as a separate, well-scoped skill.
• Skills stay minimal. Every SKILL.md is written terse on purpose — loading one costs less context than a long prompt.

See it (Before / After)

paleo compresses delivery, not meaning. Code, commands, and technical terms stay byte-exact.

[!NOTE] paleo keeps technical accuracy at 100% — it drops filler, not facts.

Features

• Modular & composable — load one skill or all seven, mix per task.
• Output + context savings — ~50–70% fewer output tokens (median ~54% on a 6-task sample — see BENCHMARK.md), plus proactive context trimming.
• Auto-detect — paleo-auto watches session state and enables the right skills automatically. No thinking required.
• Production-safe — compresses output and context only; never rewrites your code.
• Hard token budget — paleo-budget caps spend and summarizes the tail.
• Cross-agent — open Agent Skills standard: Claude Code, Codex, Gemini CLI, Qwen Code, OpenCode, Cursor, GitHub Copilot, Cline, Windsurf + 40+ agents via npx skills add.
• Zero-setup triggers — plain English phrases, no slash commands to register.
• Low overhead — each SKILL.md is intentionally terse, so loading stays cheap.
• Open & extensible — drop in your own token-saving skills.

Skills

Recommended combos

paleo is the base — keep it on. Layer the rest by what you're doing:

[!TIP] You rarely need every skill at once. paleo + paleo-auto is the default for most users — auto-detect handles the rest. For manual control: paleo + paleo-trim-context is the baseline; add paleo-converse for messy chats, paleo-summary / paleo-json for heavy tool output, and paleo-budget only when a hard cap is required.

Quick Start

# 1. Clone the collection
git clone https://github.com/mocasus/paleo.git

# 2. Claude Code — one plugin bundles all 7 skills
claude plugin marketplace add https://github.com/mocasus/paleo
claude plugin install paleo@paleo

# 3. Any agent via the open Agent Skills registry (installs to 40+ clients)
npx skills add mocasus/paleo

Then just talk to your agent — no command to register:

paleo mode · save tokens · budget 2000 · trim context

Benchmarks

Real, reproducible numbers — not hand-waved claims.

Full method, per-task table, and the runnable harness are in BENCHMARK.md. Rerun on your own stack:

export IDROUTER_API_KEY=your_key
python3 bench/benchmark.py --model claude-sonnet-4.5

[!TIP] Savings are task-dependent: biggest on verbose generative work (code, walkthroughs, comparisons — 54–79%), smaller on already-compact factual answers. paleo also cuts context tokens via paleo-trim-context, a layer a terse-persona prompt cannot reach.

Comparison

paleo is often compared with two other token-saving approachesa terse-persona system prompt and Ponytail (a code-reuse coding skill). Here is how they differ.

[!TIP] They're complementary, not rivals. Ponytail cuts the code you have to write; paleo cuts the tokens in the conversation. Terse-persona prompts proved a terse prompt helps output — paleo takes that same idea and makes it modular, adds context-trimming and a hard budget, and drops the persona gimmick. Use Ponytail for code-heavy work and paleo for chatty, long sessions.

FAQ

Tips & Triggers

paleo activates from natural-language triggers — no slash command to register. Type the trigger, the skill loads and applies.

Installation

claude plugin marketplace add https://github.com/mocasus/paleo
claude plugin install paleo@paleo

npx skills add mocasus/paleo

mkdir -p ~/.gemini/skills && cp -r skills/* ~/.gemini/skills/

hermes skills install mocasus/paleo
# Or copy skills manually
cp -r skills/paleo* ~/.hermes/skills/

npx skills add mocasus/paleo

npx skills add mocasus/paleo

npx skills add mocasus/paleo

git clone https://github.com/mocasus/paleo.git
# copy skills/paleo*/ into your agent's skills directory

All 7 skills load automatically — paleo, paleo-trim-context, paleo-auto, paleo-budget, paleo-converse, paleo-summary, paleo-json.

Full per-agent steps in INSTALL.md. See real compression numbers in BENCHMARK.md.

Hermes Integration

paleo is battle-tested on Hermes Agent by @mocasus — first user & case study.

# Install via Hermes skills manager
hermes skills install mocasus/paleo

# Or by path
cp -r skills/paleo* ~/.hermes/skills/

Then in your Hermes chat (Telegram, WhatsApp, etc.):

> paleo
🦴 paleo full — terse output, code-first

> build a REST API with FastAPI
[terse, code-first response — no preamble, no filler]

> paleo-auto
🦴 paleo-auto: watching session... enabled paleo + trim-context (23 turns)

> budget 2000
🦴 paleo-budget: 2000 output tokens, hard mode

Hermes tips:

• Start with paleo — instant token savings on every reply.
• paleo-auto for sessions >15 turns (watches context fill & enables the right skills).
• Combine paleo + budget for expensive models via provider routing.
• paleo-converse kicks in when your agent loop hits 60%+ context cap.

Custom Skills

paleo is open — wire your own token-saving skills:

1. skills/<your-name>/SKILL.md with name + description frontmatter.
2. Add the skills directory to .claude-plugin/plugin.json — the skills field is a path string (e.g. "./skills/"), not an array. Gemini + other agents pick skills up natively; no extra manifest needed.
3. Bump version badge (this file + footer) + plugin version.
4. Commit + push.

No repo edit needed — just drop any SKILL.md into your agent's skills dir (e.g. ~/.hermes/skills/<name>/). paleo loads whatever it finds under skills/.

User Stats

Share your numbers, get listed. PR your monthly token savings to this table.

How to measure:

1. Use agent 1 week without paleo → note token usage from provider dashboard
2. Enable paleo (or paleo-auto) 1 week → note new usage
3. Diff × 4 = estimated monthly savings

Contributing

Contributions are welcome — new token-saving skills, better triggers, or benchmark data.

• Open an issue describing the skill or improvement.
• Keep SKILL.md files terse (they load into context).
• Add name + description frontmatter and register in both plugin manifests.
• Bump the version badge and version fields before opening a PR.

License

MIT — see LICENSE.