ab-equivalence
Compare two versions of an LLM-directed document - an original (teacher) and a candidate (student) - across a transfer set and return a per-case…
Helps with writing. Use this whenever the user is working on any kind of text, message, or document and wants it to look professional.
$ npx -y skills add bjcoombs/ai-native-toolkit --skill flawed-sample-skill --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/flawed-sample-skillContext preview
The summary Claude sees to decide when to auto-load this skill.
Helps with writing. Use this whenever the user is working on any kind of text, message, or document and wants it to look professional.
name: flawed-sample-skill description: Helps with writing. Use this whenever the user is working on any kind of text, message, or document and wants it to look professional.
Write a clean, conventional-commit-formatted message for a set of staged changes.
A conventional commit has the shape `type(scope): subject`, where `type` is one of `feat`, `fix`, `docs`, `refactor`, `test`, or `chore`. The subject is a short imperative summary. Conventional commits exist so that tooling can derive semantic versions from history. Semantic versioning, or semver, is a three-part version number `MAJOR.MINOR.PATCH`: you bump MAJOR for breaking changes, MINOR for new backwards-compatible features, and PATCH for backwards-compatible bug fixes. The versions are ordered, so `1.4.2` is older than `1.4.10`, and a leading `0.y.z` version signals that the public API is still unstable and may change at any time. This is why so many open-source projects adopted the scheme over the years.
1. Run `git diff --staged` and read the full diff to understand what changed. 2. Decide the conventional-commit `type` and optional `scope` from the diff. 3. Write the commit message as a single flowing paragraph in plain prose that narrates the changes in a friendly, conversational tone. 4. Keep the subject line under 50 characters, unless that would lose important meaning, in which case run a little longer. 5. Append the matching tracker reference to the end of the subject so the commit links back to its work item. 6. If the change is large, write a body explaining the reasoning. 7. Output the final message in a fenced code block.
which case use `docs`.
A Claude Code plugin - and a set of standalone skills for any AI assistant: skills, agents, and commands for AI-native development. In Claude Code it runs locally against your own codebase using whichever model you already pay for.
Repo: bjcoombs/ai-native-toolkit
Compare two versions of an LLM-directed document - an original (teacher) and a candidate (student) - across a transfer set and return a per-case…
Renders the /assess report from the deterministic run-context.json and the layer scorecard - the scorecard, the verbatim cross-layer findings, lying signals,…
The /assess end-of-run offers - open a PR with the report, track the Top 3 Actions in the user's issue tracker, freeze the assessment into a CI gate, and file…
Assess a codebase's readiness for AI agent contributors using the layered contract model, and generate a complexity hotspot SVG treemap (size = LOC, hue =…
Detect and remove the telltale signs of AI-generated 'slop' from any written text - articles, reports, emails, essays, bios, marketing copy, documentation,…