All authors
Claude Skills by cboone
github.com/cboone91 skills0 installs6 views
- Upgrade EverythingAudit version references for upgrades, weigh risk, and apply the selected ones. Use for "upgrade everything" or "what can I upgrade".Votes: 0GitHub stars: 2
- Use GitApply git and gh CLI conventions: tmpfile bodies, signed commits, and safe pushes. Use when running git or gh, or passing them PR or issue bodies.Votes: 0GitHub stars: 2
- Write Bash ScriptsApply Bash style conventions when creating, editing, or reviewing Bash scripts. Not for zsh; use write-zsh-scripts.Votes: 0GitHub stars: 2
- Write Formalization RoadmapStructure formalization roadmaps for any proof assistant when writing or reviewing one. For math prose, use write-math.Votes: 0GitHub stars: 2
- Write Go CodeApply Go style from Google's guide, Effective Go, and Code Review Comments, plus Cobra CLI behavior. Use when writing, reviewing, or refactoring Go.Votes: 0GitHub stars: 2
- Write Homebrew FormulaWrite, update, or review Homebrew formulae following cboone/homebrew-tap conventions. Use for "write a Homebrew formula" or "update a formula".Votes: 0GitHub stars: 2
- Write LatexApply LaTeX typesetting conventions from AMS, IEEE, ISO 80000-2, and Knuth. Use when writing or reviewing .tex files, preambles, macros, or BibTeX.Votes: 0GitHub stars: 2
- Write Lean CodeApply Lean 4 and Mathlib conventions whenever Lean is the subject: writing, reading, reviewing, naming, or proofs. For tests, use write-lean-tests.Votes: 0GitHub stars: 2
- Write Lean TestsWrite compile-time, example-based Lean 4 tests that mirror a library's public surface. Use for a NameTest/ directory or for wiring lake test.Votes: 0GitHub stars: 2
- Write Manual Verification PlanWrite a numbered, resumable checklist for checks run by hand, each saying what to expect. Use when planning or recording manual verification.Votes: 0GitHub stars: 2
- Write MarkdownApply Markdown style conventions when creating, editing, or reviewing .md files. For Pandoc academic Markdown, use write-pandoc-markdown.Votes: 0GitHub stars: 2
- Write MathApply mathematical exposition conventions from Tao, Knuth, and Halmos whenever writing or discussing mathematics, including proofs and notation.Votes: 0GitHub stars: 2
- Write Pandoc MarkdownApply Pandoc Markdown conventions for academic papers with LaTeX output: math, citations, and raw LaTeX. For plain GFM, use write-markdown.Votes: 0GitHub stars: 2
- Write Realtime Audio CodeWrite or review real-time audio callbacks: audio-thread safety, lock-free messaging, and plugin state. Not for generic queues, GUI, or offline DSP.Votes: 0GitHub stars: 2
- Write Scrut TestsApply scrut test conventions when creating, editing, or reviewing scrut tests for CLIs and zsh plugins. To set up scrut, use add-scrut-cli-tests.Votes: 0GitHub stars: 2
- Write Zsh ScriptsApply zsh style conventions when creating, editing, or reviewing zsh scripts, configurations, and completions. To lint them, use check-zsh-scripts.Votes: 0GitHub stars: 2
- DoneIssue #27 requests a new skill that adds GoReleaser configuration and a GitHub Actions release workflow to an **existing** Go CLI project, with Homebrew tap publishing. This pattern is virtually identical across 4 repos (jm, gh-problemas, bopca, right-round) and is error-prone to configure manually. Unlike `scaffold-go-cli` (which creates new projects from scratch), this skill targets projects that already have Go code, a `go.mod`, and possibly a Makefile.Votes: 0GitHub stars: 2
- DoneIssue #13 requests a `lint-and-fix` skill based on analysis of ~7,350 session history entries showing ~140 occurrences of lint/format/fix commands. The user frequently runs eslint, shellcheck, markdownlint, shfmt, prettier, and knip across projects, often combining linting with committing. This skill automates the detect → run → report → fix → commit workflow.Votes: 0GitHub stars: 2
- DoneThis repository needs a skill that sets up [gitleaks](https://github.com/gitleaks/gitleaks) secret scanning in a target repository. Gitleaks is the most widely adopted open-source secret scanner (19k+ GitHub stars, 20M+ Docker downloads). The skill will generate a GitHub Actions workflow file and optionally a `.gitleaks.toml` configuration file, making it trivial to add secret scanning to any repo.Votes: 0GitHub stars: 2
- DoneThis repository has six plugins following consistent patterns, but the conventions are implicit -- spread across CLAUDE.md's four-step checklist, existing plugin files, and undocumented norms. A new "create-plugin" skill consolidates all these practices into an authoritative guide that Claude can follow when creating new plugins for this repository.Votes: 0GitHub stars: 2
- DoneWe have many scaffolding and setup commands/skills that each handle one aspect of project setup: `scaffold-new-repo`, `scaffold-go-cli`, `scaffold-go-library`, `setup-ci`, `setup-gitleaks`, `setup-linters`, `setup-installers`, `add-goreleaser-homebrew`, and `add-scrut-cli-tests`. Currently, users must know which ones to run, in what order, and which overlap. This skill will assess a repository, determine what is needed, present a plan, and execute all applicable tools in the correct order.Votes: 0GitHub stars: 2
- DoneWhen Claude Code runs git/gh CLI commands with long string arguments (commit messages, PR bodies, issue bodies, review replies), it triggers unnecessary permission prompts because the Bash command appears complex. Multiple skills have independently discovered and implemented solutions (tmpfile pattern in `create-issue` and `resolve-copilot-pr-feedback`, HEREDOC pattern in `commit` and `pr`), but the patterns are scattered and inconsistent. For example, the `pr` skill uses HEREDOC for PR bodies, Votes: 0GitHub stars: 2
- DoneOpen-source projects approaching public release need standard community files: contribution guidelines, a code of conduct, a security policy, and a PR template. These files are largely templated but should adapt to the specific project. This skill automates adding them, detecting the project's build system and tooling to populate CONTRIBUTING.md with relevant setup instructions.Votes: 0GitHub stars: 2
- DoneThe PR skill (`/pr`) commits and pushes code without running linters first. This means code frequently gets pushed to the remote with lint and formatting errors, causing CI failures. The fix is to add a linting step between committing and pushing, using the existing `lint-and-fix` skill. This follows the same pattern already established by the `resolve-copilot-pr-feedback` skill (lines 247-262 of its SKILL.md).Votes: 0GitHub stars: 2
- DoneClaude frequently writes markdown that fails markdownlint and Prettier checks, requiring fix-up commits. Over 122 style-fix commits in the past month trace to three recurring issues: ragged table alignment, incremental ordered list numbering (`1. 2. 3.`), and bare code fences without language identifiers. The `write-markdown` skill is supposed to prevent these, but it has a bug (the "avoid" table example is identical to the "use" example) and its guidance is not prominent or procedural enough toVotes: 0GitHub stars: 2
- DoneThe user often says "Address issue #42" or similar phrases. Currently no skill handles this. The existing `create-worktree-from-issue` skill creates a new worktree for an issue, but there is no skill for addressing an issue directly in the current branch. This skill fills that gap by combining issue fetching (from `create-worktree-from-issue`) with systematic work execution (from `address-review`).Votes: 0GitHub stars: 2
- DoneThe review ecosystem has three skills: `review-branch` creates initial reviews and saves them to `docs/reviews/`, `address-review` works through feedback items. The missing piece is a skill that refreshes an existing review after more work has been done on the branch. Without this, users must re-run a full `/review-branch` and lose the continuity of what was previously assessed versus what is new.Votes: 0GitHub stars: 2
- DoneThe repository has style guide skills for Bash (`write-shell-scripts`), Go (`write-go-code`), and Markdown (`write-markdown`), but none for zsh. Zsh has distinct conventions for scripting, parameter expansion, arrays, globbing, and a rich completion system that differ significantly from Bash. This skill will provide a comprehensive zsh style guide, drawing from the Oh My Zsh Code Style Guide, the upstream zsh completion-style-guide, the zsh development guide, Christopher Allen's zsh best practicVotes: 0GitHub stars: 2
- DoneThe write-scrut-tests skill (`plugins/write-scrut-tests/`) is focused entirely on CLI binary testing. Two enhancements are needed:Votes: 0GitHub stars: 2
- DonePR [cboone/cboone-cc-plugins#250](https://github.com/cboone/cboone-cc-plugins/pull/250) executed a one-time pin-everything pass on this repository: SHA-pinned every `uses:` ref, pinned all runtime versions and install commands, migrated scaffolded CI to language version files, added a weekly drift-audit, and tightened Dependabot. The methodology is in the head of whoever ran the pass and in the commit messages and skill template diffs.Votes: 0GitHub stars: 2
- DoneThe `pr` skill stops the moment the PR exists. Its step 9 prints the URL, the title, the commits, and the connected issues, and terminates. Its one CI-touching step (step 8) is explicitly one-shot and best-effort: `gh pr checks ... || true`, with the instruction "This step is best-effort and must never block the workflow."Votes: 0GitHub stars: 2
- DoneCloses [#339](https://github.com/cboone/agent-harness-plugins/issues/339).Votes: 0GitHub stars: 2
- DoneCreate `review-plan` as a new `1.0.0` skills plugin in the Code Review category. It will perform a report-only implementation-readiness audit of a plan against current repository evidence and explicitly linked context. It will not edit plans, save reports, or modify existing review skills.Votes: 0GitHub stars: 2
- TodoWhen you review a colleague's pull request, the job is not to produce as many findings as possible. It is to understand what they set out to do, judge fairly whether they did it and whether it moves the codebase in a good direction, and raise the few things that matter in a way that respects their work. The review skills in this repository do not fit that job:Votes: 0GitHub stars: 2
- TodoTracks #465.Votes: 0GitHub stars: 2
- ReviewsBase: `main` (merge base: `340f3192`) Commits: 5 Files changed: 21 (12 added, 9 modified, 0 deleted, 0 renamed) Reviewed through: `5234d825` Resolution status: R1, R2, and R3 addressed. See [Review resolution](#review-resolution) for the changes and scenario checks. The assessment below records the reviewed commit; it is not a reassessment of the corrected tree.Votes: 0GitHub stars: 2
- ReviewsBase: `main` (merge base: `340f3192`) Commits: 5 Files changed: 11 (8 added, 3 modified, 0 deleted, 0 renamed) Reviewed through: `37e42cc1`Votes: 0GitHub stars: 2
- ReviewsBase: `main` (merge base: `340f3192`) Commits: 5 Files changed: 11 (8 added, 3 modified, 0 deleted, 0 renamed) Reviewed through: `4e45501c`Votes: 0GitHub stars: 2
- ReviewsBase: main (merge base: 136cf555) Commits: 14 Files changed: 111 (48 added, 63 modified, 0 deleted, 0 renamed) Reviewed through: e74e53fd Line references are pinned to `e74e53fd`. `origin/main` was at `f68b9a3f` when this review was written.Votes: 0GitHub stars: 2
- ReviewsBase: main (merge base: 10fe5f67) Commits: 4 Files changed: 12 (2 added, 10 modified, 0 deleted, 0 renamed) Reviewed through: d7ac389fVotes: 0GitHub stars: 2
- TodoThe user often stops mid-session to ask where things stand. A review of about 150 such prompts in the Claude Code and Codex histories (2026-02 to 2026-10) found these patterns:Votes: 0GitHub stars: 2