agentleFS
Sign inStart free
0n1-Studios0n1-Studios/marginalia/CLAUDE.md

marginalia

CLAUDE.md1 starsChanged 52 days agoMITView on GitHub

A Claude Code skill that turns screenshots into a published, annotatable review board: the user clicks a screenshot, a post-it lands at that point, they type, and any later session reads the notes back as structured input keyed to board + x/y percentage. The point is that position is the feedback — prose is a lossy channel for "the third card". This repo and the installed skill at ~/.claude/skills/marginalia/ are mirrors. A change to either is owed to the other…

The file

We split this CLAUDE.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 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.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Only SKILL.md and assets/ are mirrored.

    line 20
  2. Distinctive:

    No absolute paths, hostnames, local ports, or internal service names in anything that ships — this file included.

    line 28
  3. Very specific:

    No screenshots of the user's own projects in docs/ unless they are meant to be public — docs/board.jpg is the deliberate exception.

    line 31

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Marginalia

A Claude Code skill that turns screenshots into a published, annotatable review board: the user clicks a screenshot, a post-it lands at that point, they type, and any later session reads the notes back as structured input keyed to board + x/y percentage. The point is that position is the feedback — prose is a lossy channel for "the third card".

The two-copy rule (load-bearing)

This repo and the installed skill at ~/.claude/skills/marginalia/ are mirrors. A change to either is owed to the other — they drift silently otherwise, and the installed copy is the one that actually runs.

diff -r ~/Development/Marginalia ~/.claude/skills/marginalia \
  --exclude=.git --exclude=docs --exclude='*.md' --exclude=LICENSE

Only SKILL.md and assets/ are mirrored; README.md, LICENSE and docs/ are repo-only.

This is the studio's first PUBLIC repo

0n1-Studios/marginalia, MIT. Everything committed here is world-readable, so it holds a different bar from the rest of the workspace:

  • No absolute paths, hostnames, local ports, or internal service names in anything that ships — this file included. The skill must work for someone outside the studio, on a machine that runs none of its other software.
  • No screenshots of the user's own projects in docs/ unless they are meant to be public — docs/board.jpg is the deliberate exception.
  • README is the front door for strangers; keep the install path copy-pasteable.

Layout

FileWhat it is
SKILL.mdThe skill — build procedure, read-back parse, the update-without-losing-notes rule
assets/build.pyManifest → self-contained HTML. Inlines screenshots as data URIs, fails loudly if the page exceeds the artifact budget
assets/template.htmlThe board: post-it render, drag, edit, flag, ledger, save-as-republish. No dependencies beyond Google Fonts

Footguns

  • Rebuilding overwrites saved notes. A refresh or a new board must pull the published page's current state and rebuild with --state state.json. This is the one way to lose a user's work here.
  • Only the owner can save. An unauthenticated viewer still renders and reads the board, but cannot publish — the page says so rather than failing silently.
  • The page must stay self-contained. Artifact CSP blocks every external host except Google Fonts, so screenshots ride as data URIs and count against the 16MB budget. build.py enforces this; don't route around it.

Roadmap

Open work lives in TODO.md — delete an item's line when it ships; bugs go to GitHub issues (label code-review), never the roadmap.

Conversation Title

MARGINALIA: brief description

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free