llm-skills

KO | EN

gitlite — search

#hacktoberfest #machine-learning #rust #linux #macos #java #windows #awesome #kubernetes #awesome-list #pytorch #mcp #openai #open-source

(★ 9)

Pony related LLM skills

llm-skills Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Pony LLM Skills

Skills for working with [Pony](https://www.ponylang.io) in any LLM coding harness that supports sub-agents. Each skill is a self-contained reference that your LLM loads on demand during coding sessions.

## Tested with

- [Claude Code](https://docs.anthropic.com/en/docs/claude-code)

## Compatible with

- [OpenAI Codex](https://developers.openai.com/codex) — Codex loads the same `SKILL.md` format from `~/.agents/skills`, where `install.py` installs them.

**About the `pony-` prefix:** All skills in this repo use a `pony-` prefix as an org namespace to avoid name collisions with skills from other sources. Some skills (like `pony-ref`) are Pony-language-specific. Others (like `pony-ensemble` and `pony-code-review`) are language-agnostic methodology skills that work on any codebase — the prefix is about where they come from, not what languages they apply to.

## Installation

Clone the repo and run the install script. It detects which harnesses you have installed and symlinks each skill into their skills directory, so the skills stay up to date when you pull.

```bash
git clone https://github.com/ponylang/llm-skills.git
cd llm-skills
python install.py
```

With no arguments, `install.py` installs for every harness it detects: Claude Code (into `~/.claude/skills/`) and Codex (into `~/.agents/skills/`). To target a specific harness regardless of what's detected, pass `--claude` and/or `--codex`.

That's it. Start a new session and the skills are available.

**Invoking a skill:** In Claude Code, load a skill with a slash command like `/pony-ref`. In Codex, skills load automatically based on their description, or you can mention one explicitly with `$pony-ref`.

To update later:

```bash
cd llm-skills
git pull
```

No re-install needed — the symlinks point to your clone, so pulling new content updates the skills automatically.

### Uninstalling

```bash
python install.py --uninstall
```

This removes the symlinks from the detected harnesses (or from the harness you select with `--claude`/`--codex`), not the cloned repo.

## Skills

### Routing

#### pony-skills

A routing index for the other skills — load it (or reference it from a Pony project's `CLAUDE.md` or `AGENTS.md`) and it tells you which `pony-*` skill to load for each task. The single-trigger alternative to wiring up each skill's trigger by hand; a good place to start.

### Pony Language

#### pony-ref

The Pony language reference. Load it at the start of a Pony coding session or when you hit a question about capabilities, the type system, the runtime, or testing.

What's in the quick reference (loaded into context automatically):

- Reference capabilities table, subtyping rules, and key patterns (consume, recover, destructive read)
- Common gotchas (iso aliasing, async stdin, scoping, type aliases)
- Integer arithmetic modes (wrapping, partial, checked)
- Syntax essentials
- PonyCheck property-based testing patterns and gotchas
- Stdlib pitfalls (Reader, Array, Writer, buffered)
- Panic primitives ("mort" pattern)

What's in the `references/` directory (read on demand for deeper questions):

- **Type system synopsis** — distilled from the academic papers. Deny-properties matrix, subtyping lattice, viewpoint adaptation tables, safe-to-write rules, recovery, generics.
- **Runtime/GC synopsis** — ORCA object GC, MAC actor cycle collection, per-actor heaps, causal messaging, the scheduler. Includes an "Implementation Divergences" section documenting where the current ponyc runtime has evolved beyond the papers.
- **Academic papers** — the full text of all nine Pony papers covering the type system, garbage collection, generics, and distributed programming.
- **Website content** — snapshots of the Pony tutorial, patterns cookbook, and main website (via their llms.txt files). Covers language fundamentals, idiomatic patterns, tooling guides, and FAQ.

### Project Conventions

#### pony-prose

How the words read in any prose that ships with the code — comments, docstrings, release notes, READMEs, a project's `AGENTS.md`, commit messages. Load it before writing any of them. It is the layer under the form-specific skills: they say what belongs in each kind of prose, this says how to write it plainly.