claude-skills
CLAUDE.md1 starsChanged 37 days agoMITView on GitHub
This repo is the bendyworks Claude Code plugin marketplace: the skills, guidance files, and CLIs other people install and run. Contributor mechanics -- how changes ship, the checks to run before a PR, how to write and dry-run a skill -- live in CONTRIBUTING.md. Read it before opening a PR; this file carries only what a session working here needs in order to behave correctly. It was formerly bendyworks/claude-skills. Never create a repository by that name in the bendyworks org:…
In this repository15 other files it gives its agents.
- app-wind-downSkill
- architecture-surveySkill
- bug-cluster-ledgerSkill
- change-highlightsSkill
- dependabot-batchSkill
- finished-issue-housekeepingSkill
- gauntletSkill
- linearSkill
- markdown-to-pdfSkill
- parallel-checkoutsSkill
- pause-workSkill
- plan-issueSkill
- resume-workSkill
- review-open-itemsSkill
- targeted-specsSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 17
Every issue, rule, and skill here is weighed against these goals.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
rules-that-bend
This repo is the bendyworks Claude Code plugin marketplace: the skills,
guidance files, and CLIs other people install and run. Contributor mechanics
-- how changes ship, the checks to run before a PR, how to write and
dry-run a skill -- live in CONTRIBUTING.md. Read it before
opening a PR; this file carries only what a session working here needs in
order to behave correctly.
It was formerly bendyworks/claude-skills. Never create a repository by
that name in the bendyworks org: GitHub redirects it here only while the
name is free, and existing installers fetch the marketplace through that
redirect.
Goals
Every issue, rule, and skill here is weighed against these goals. A good idea that serves none of them is put to the maintainer as a question, saying so, and is not filed or built until the maintainer answers.
- An elegant, efficient open source project. A reader can copy one skill or one guidance file without taking the rest, learn from how it is written, and judge the maintainer's work by it. A skill someone can read in one sitting is worth more than one that covers every case.
- Fast checks. The full suite runs in under three minutes, and a gauntlet run is short enough that nobody plans around it. A test or a check that slows either one has to be worth the time it adds to every later story.
- A shrinking issue list. More issues close than open. A follow-up is filed with care and implemented straight away, in the story that found it or directly after. An idea not worth doing soon is dropped, not filed.
- Less, where less is better. Work that cost more than it gave is found, then simplified or removed. Removing something is an ordinary outcome of a story.
How the work is divided
The maintainer and a session write the issue and the plan together. The session then implements the plan alone: it makes the development decisions, writes and squashes the commits, and writes the pull request, title and body included. The maintainer reviews that pull request closely, and the review is where the session's decisions get overruled.
A session's question about a development detail works against this division. So does a pull request the maintainer has to rewrite before reviewing it: the pull request is the session's deliverable, and its quality is judged as such.
Deploy-on-Merge Mode
This project uses Deploy-on-Merge Mode: merging to the default branch is the production deploy, with no later step that could still fail or be skipped.
The plugin carries no version field, so the merged commit is what installers
get. An installer running /plugin marketplace update bendyworks later is
adoption, not a deploy step that could fail, so nothing stands between a merge
and the change being live.
Merging here is the maintainer's, not a session's: this project declares no Session-Merge Mode, so a session prepares a PR and hands it off. That matters more here than on most repos, because the merge it would be claiming is the production deploy above.
The checks.yml workflow that runs on pushes to main is not a step in the
way: it verifies what installers already have and cannot withhold a commit
from them, so a red run there is a bug to fix, not a deploy that failed.
Tracker
GitHub Issues is this repo's tracker of record. GitHub issues here are called RTB-NNN: RTB-NNN is issue #NNN.
Older references say CS-NNN, the key from when this repo was named claude-skills. CS-NNN is GitHub issue #NNN, never a Linear ID. Names that already say CS keep it. Nothing new is named with CS, whatever the issue's number: new branches, plan files, and pull request titles use RTB, and so does new work on an issue that shipped under CS.
Everything here is public
This repo is public and its history is permanent. No client names, other
projects' tracker IDs (use the neutral ABC-NNN form), personal email
addresses, or absolute home-directory paths -- in code, prose, commits, or PR
descriptions. This repo's own RTB-NNN, the older CS-NNN, and #NNN are
fine in commits, PR titles, and PR descriptions; shipped skills and guidance
still use placeholders.
CONTRIBUTING.md has the full rule and the checks that enforce the mechanical
half of it.
Skills and guidance stay installer-generic: they are read by people with no access to our machines or our private notes, so they must not reference personal skills, private CLAUDE.md files, or anything a reader outside Bendyworks could not act on.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

