claudeers.

🔓 unclaimed — this page was auto-generated from GitHub. Are you the creator?

Claim this page →
// Claude Skills

prompt-warrior

Agent skill: your AI coding logs as a gamified character sheet — Claude Code, Codex, OpenCode, Gemini, Copilot. 74 achievements, fixed-scale metrics, LLM-wri…

// install
git clone https://github.com/timoncool/prompt-warrior
Prompt Warrior emblem

Prompt Warrior

Your AI agent's logs are hiding a character sheet. We found it.

English · Русский

Prompt Warrior card

Prompt Warrior is an agent skill that reads your local session logs — Claude Code out of the box; Codex CLI, OpenCode, Gemini CLI and Copilot via documented log formats — and turns them into something between a psychological profile and an RPG character sheet: real numbers on a fixed scale, a title you didn't choose but absolutely deserve, and achievements you've been grinding for months without knowing it. Runs 100% locally on stdlib Python — nothing ever leaves your machine.

Why bother?

  • You love achievements, damn it. 74 of them, common to legendary, Steam-style cards with rarity colors and hover-to-reveal unlock conditions. You already earned some of them — you just haven't been told yet.
  • You'll finally learn who you are. Not "a developer" — a Furious Terminal Commander, level 90, "The Iron Fist". Class comes from your harness, race from your favorite model: a Claude Code Warrior riding a Mighty Opus.
  • The AI writes your chronicle. A short grimoire-flavored biography composed from your actual behavior. It's weirdly accurate and slightly uncomfortable. In a good way.
  • Numbers nobody else will show you. Your werewolf index (night profanity vs day), session boiling point (how many messages until the first flare-up), double-texting rate, subagent armies spawned, the exact hour you are most dangerous.
  • A fixed scale means bragging rights. Formulas are frozen (SCALE). Same math for everyone — swap cards with a friend and settle who's the real Tyrant.
  • It roasts you with receipts. Strengths and weaknesses derived strictly from your own metrics, every advice backed by a cited study — debunked prompting myths listed too (sources.md).
  • Your monster evolves. A deterministic creature is assembled offline from your title (optional robohash package, zero network) — change your ways, change your monster.
  • 100% local and private. stdlib-only Python, zero dependencies, no API keys, no network calls. Your logs, your machine — your shame stays yours.

A taste of the achievements

AchievementRarityUnlock
Tyrant🟡 legendarynegativity outnumbers praise 10:1
Gutenberg🟡 legendary25M+ tokens generated
Werewolf🟣 epicnight profanity ≥ 1.5× the daytime dose
Short Fuse🟣 epicfirst flare-up by message three of a session
Night's Watch🔵 rare30%+ of activity between midnight and 6 a.m.
Cowboy⚪ commonzero plan modes across 50+ sessions

…and 68 more, from Sprinter to Lord of Legions. Conditions are honest, thresholds are frozen, nobody gets one for free.

What's on the card

  • Title & identity — epithet + rank + level, class and race, monster avatar
  • Warrior's Chronicle — the LLM-written mini-biography
  • Message volume, top imperatives, tone markers — how you actually talk
  • Hours & days of week — when you actually work
  • Six gauges — Rage, Warmth, Rigor, Nocturnality, Impatience, Cache Thrift
  • Economy & arsenal — tokens burned (deduplicated like ccusage), cache efficiency, tool roles (Operator / Surgeon / Archaeologist / Puppeteer), MCP fleet, file extensions you actually touch, most expensive sessions
  • Battle habits — 12 deep-cut signals from interruptions to RU/EN code-switching
  • Signature vocabulary — the words that are unmistakably yours
  • Strong & weak — with a fix and a source for every weakness
  • Progress between visits — come back in a week: new achievements, level-ups, metric shifts. That's the share-worthy moment.

Everything ships in both RU and EN, symmetrically — the card speaks your language. Cross-harness aware: session-log formats for Codex CLI, OpenCode, Gemini CLI and Copilot are documented in references/harnesses.md.

The inline widget

The same card renders as a live widget right in the chat — with an "Open HTML card" button, hover unlock conditions and the achievements accordion:

Inline widget — title, chronicle, gauges Inline widget — arsenal, habits, achievements

Quick Start

The lazy way — one message. Paste this into Claude Code and it installs and runs everything by itself:

Install the Prompt Warrior skill and run it end to end.

Install: git clone https://github.com/timoncool/prompt-warrior into YOUR OWN
skills directory, named prompt-warrior (for Claude Code that is
~/.claude/skills/; other harnesses — wherever your skills live).

Then follow the installed skill's SKILL.md as a regular user: all-time profile,
HTML card + open it in the browser, inline widget, profile breakdown and
recommendations. Don't invent anything beyond the skill.

Or step by step:

  1. Clone

    git clone https://github.com/timoncool/prompt-warrior ~/.claude/skills/prompt-warrior
    

    Windows (PowerShell):

    git clone https://github.com/timoncool/prompt-warrior "$env:USERPROFILE\.claude\skills\prompt-warrior"
    

    or install as a plugin:

    /plugin marketplace add timoncool/prompt-warrior
    /plugin install prompt-warrior@prompt-warrior
    
  2. Ask Claude Code

    /prompt-warrior
    

    or just say: "build my prompt warrior profile".

  3. Get your card — an inline widget plus a self-contained ai-profile.html you can open, screenshot and share.

Usage

  • "build my prompt warrior profile" — all-time profile
  • "my prompt profile for the last week" — last 7 days
  • "профиль за июнь" — exact date range
  • "profile for project X" — a single project (--project)
  • The period is always your choice; the skill never picks one silently.
  • No browser, no widget? The card falls back to plain text right in the console.

Under the hood: scripts/analyze.py reads ~/.claude/projects JSONL logs (read-only), deduplicates your messages, computes metrics and writes profile.json; Claude authors only the chronicle and the strengths/weaknesses picks (content.json), and scripts/render.py builds the whole card from them — the model never hand-writes HTML or reads SVG.

Documentation

Other Projects by @timoncool

ProjectDescription
telegram-api-mcpFull Telegram Bot API as MCP server
civitai-mcp-ultimateCivitai API as MCP server
trail-specTRAIL — cross-MCP content tracking protocol
GitLifeYour life in weeks — interactive calendar
ACE-Step StudioAI music studio — songs, vocals, covers, videos
VideoSOSAI video production in the browser

Authors

Support the Author

I build open-source software and do AI research. Most of what I create is free and available to everyone. Your donations help me keep creating without worrying about where the next meal comes from =)

All donation methods | dalink.to/nerual_dreming | boosty.to/neuro_art

  • BTC: 1E7dHL22RpyhJGVpcvKdbyZgksSYkYeEBC
  • ETH (ERC20): 0xb5db65adf478983186d4897ba92fe2c25c594a0c
  • USDT (TRC20): TQST9Lp2TjK6FiVkn4fwfGUee7NmkxEE7C

And if the card made you smirk — star the repo ★. Think of it as an achievement for the maintainer.

Star History

Star History Chart

License

Code: MIT. Achievement, section and emblem icons: game-icons.net authors, CC-BY-3.0 (attribution). Monster art (set2) by Hrvoje Novakovic, CC-BY-3.0, assembled locally by the MIT-licensed robohash package.

// compatibility

Platformscli, api, web
Operating systems
AI compatibilityclaude
LicenseMIT
Pricingopen-source
LanguagePython

// faq

What is prompt-warrior?

Agent skill: your AI coding logs as a gamified character sheet — Claude Code, Codex, OpenCode, Gemini, Copilot. 74 achievements, fixed-scale metrics, LLM-written chronicle, research-backed advice. 100% local, RU/EN.. It is open-source on GitHub.

Is prompt-warrior free to use?

prompt-warrior is open-source under the MIT license, so it is free to use.

What category does prompt-warrior belong to?

prompt-warrior is listed under skills in the Claudeers registry of Claude-compatible tools.

0 views
14 stars
unclaimed
updated 2 days ago

// embed badge

prompt-warrior on Claudeers
[![Claudeers](https://claudeers.com/api/badge/prompt-warrior.svg)](https://claudeers.com/prompt-warrior)

// retro hit counter

prompt-warrior hit counter
[![Hits](https://claudeers.com/api/counter/prompt-warrior.svg)](https://claudeers.com/prompt-warrior)

// reviews

// guestbook

0/500

// related in Claude Skills

🔓

An agentic skills framework & software development methodology that works.

// skillsobra/Shell245,541MIT[ claude ]
🔓

Public repository for Agent Skills

// skillsanthropics/Python157,979[ claude ]
🔓

💫 Toolkit to help you get started with Spec-Driven Development

// skillsgithub/Python117,790MIT[ claude ]
🔓

AI coding assistant skill (Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and more). Turn any folder of code, SQL schemas, R scripts, shell scripts, docs,…

// skillsGraphify-Labs/Python77,228MIT[ claude ]
→ see how prompt-warrior connects across the ecosystem