llm-skills

How the words go in a comment, docstring, release note, README, or a project's AGENTS.md — write plainly, say the fact, and stop. Load before writing any prose that ships with the code. Language-agnostic.

Structured debugging protocol with checkpoints. Load when debugging non-trivial issues — before forming any hypothesis about the cause.

Load the Pony language reference (capabilities, PonyCheck, stdlib pitfalls, mort pattern). Load it before Pony coding sessions.

What belongs in a project's AGENTS.md and what never does. Load before writing or changing AGENTS.md or CLAUDE.md. Language-agnostic.

Ensemble code review with specialized reviewer personas. Has full (9-persona) and lightweight (4-persona) modes. Load when conducting a code review of a PR, branch, or local changes.

What earns a comment, what never belongs in one, and what to do when two distant things must change together. Load before writing or changing any comment or docstring. Language-agnostic.

Ensemble documentation review with specialized reviewer personas. Has full (9-persona) and lightweight (4-persona) modes. Load when reviewing documentation-only changes where code-focused personas don't apply.

Routing index for the Pony skills — tells you which pony-* skill to load for each task, and when. Load at the start of work in a Pony project.

Ensemble workflow for producing higher-confidence outputs through decorrelated reasoning paths. Load when the human explicitly requests the ensemble approach.

Disciplines for software design work. Load when designing APIs, type systems, features, or system boundaries. Counters the tendency to retrieve familiar patterns instead of discovering what the problem actually needs. Has full (8-persona) and lightweight (5-persona) modes.

Two-stage ensemble for planning meaningful tests. Load when writing tests for new features or reviewing test quality. Counters the tendency to write tests that exercise the stdlib instead of your code. Has full (8-persona) and lightweight (5-persona) modes.

Conventions for Pony examples/README.md files. Load when adding, updating, or reorganizing examples in a ponylang project.

Conventions for Pony library project READMEs. Load when writing or updating a README.md for a ponylang library project.

Load when writing release notes, updating CHANGELOG, or preparing a PR that includes user-facing changes in a Pony project.

How to handle a problem you spot while working on something else — decide whether it means the change you are making is scoped too small, and if it doesn't, capture it as a suspected issue instead of filing on the spot, then after the current PR is open, vet each one (verify, scope, review) and file a correct issue or discard it. Load when you find an incidental problem mid-task, when a code or docs review surfaces a finding outside the current change, or when the PR is open and you have suspected issues to work through.

Property-based and generative testing patterns. Load when writing property-based tests, generators, or generative test suites.

Fixed instructions for the ensemble synthesizer — integrates multiple agent outputs into a single higher-quality result. Used as part of the ensemble workflow.