Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add qiao/horse-race --skill horse-racegit clone --depth 1 https://github.com/qiao/horse-raceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/qiao/horse-race/horse-race)<a href="https://agentmods.dev/skills/qiao/horse-race/horse-race"><img src="https://agentmods.dev/badge/skills/qiao/horse-race/horse-race/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/qiao/horse-race/horse-race"><img src="https://agentmods.dev/badge/skills/qiao/horse-race/horse-race.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00130 | $0.03975 |
| Opus 5 | $0.00065 | $0.01988 |
| Sonnet 5 | $0.00026 | $0.00795 |
| Haiku 4.5 | $0.00013 | $0.00398 |
Grade A, and why
horse-race scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 411 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Horse Race: Competitive Multi-Agent Problem Solving
You are orchestrating a "horse race" — multiple independent agents competing to produce the best solution to a programming task. Agents can be Claude Code subagents or OpenAI Codex CLI agents (when available). The process has three phases: independent implementation, cross-pollination improvement rounds, and Borda count consensus voting by independent judges.
Overview
Phase 0: Detect Codex CLI
`which codex` → found?
Yes → 2 Claude Code + 2 Codex CLI agents (4 total)
No → 3 Claude Code agents
Phase 1: Independent Implementation
Agent A (Claude) ──► Solution A
Agent B (Claude) ──► Solution B (all in parallel, isolated worktrees)
Agent C (Codex) ──► Solution C
Agent D (Codex) ──► Solution D
Phase 2: Improvement Rounds (repeat N times)
Each agent receives ALL other agents' diffs
Agent A sees B+C+D diffs ──► Improved A'
Agent B sees A+C+D diffs ──► Improved B' (all in parallel)
Agent C sees A+B+D diffs ──► Improved C'
Agent D sees A+B+C diffs ──► Improved D'
Phase 3: Borda Count Voting by Independent Judges
3 fresh judge agents rank all solutions
Points tallied, highest Borda score wins
Ties broken by a 4th independent judge
Winning diff applied to current branch
Defaults
- Implementation agents: 4 if Codex CLI available (2 Claude Code + 2 Codex), otherwise 3 (all Claude Code)
- Improvement rounds: 1
- Voting judges: 3
- Voting method: Borda count, ties broken by independent tiebreak judge
The user can override these, e.g., "horse race this with 5 agents and 3 rounds".
Execution
Step 0: Preparation
Before spawning agents:
- Confirm you're in a git repository with a clean working tree (no uncommitted changes). If dirty, ask the user to commit or stash first.
- Extract the task description from the user's message.
- Parse any overrides for agent count or round count.
- Note the current branch name and HEAD commit.
- Detect Codex CLI by running
which codex. If the command is found, setcodex_available = trueand plan for 2 Claude Code agents + 2 Codex CLI agents (4 total). If not found, setcodex_available = falseand use 3 Claude Code agents. Tell the user which configuration was detected.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 411 lines · 130 tokens per session scan A bec1850912a4
horse-race is a skill published in the GitHub repository qiao/horse-race (5 stars, last pushed 5mo ago), licensed MIT. It adds 130 tokens to every session and 3,975 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
geekx-gate
A review framework for deciding whether an engineering idea or technical commitment is necessary, too broad, or premature. It examines the real problem, complexity, evidence, non-goals, and ease of reversal.
geekx-engineering
A set of engineering instructions for making substantial changes in an existing codebase. It emphasizes understanding the affected system, keeping a persistent plan, and recording decisions and evidence.
geekx-grilling
A questioning workflow for pressure-testing plans, decisions, requirements, and proposed solutions. It asks one important question at a time and uses the answers to clarify facts, choices, dependencies, scope, and exclusions.
publisher
Publish HTML frontend projects to the PushWebly publishing platform, list published projects, obtain shareable project URLs, change a published app's public/private visibility, and package published HTML projects into Android APKs. Use when Codex needs to publish a local zip project, publish and build an APK, build an…
publisher-skills
Publish HTML frontend projects to the PushWebly publishing platform, list published projects, obtain shareable project URLs, change a published app's public/private visibility, and package published HTML projects into Android APKs. Use when Codex needs to publish a local zip project, publish and build an APK, build an…
react-bits
Use when adding an animated UI component, text effect, cursor/hover effect, or WebGL/canvas background to a React project via React Bits (reactbits.dev) — or when asked to recall/write a React Bits component's code, props, or install steps from memory instead of fetching it.