{"id":"e1895f7e-3847-45a3-90a2-caf19d31d312","slug":"ymendel-dotclaude","kind":"skill","name":"dotclaude","summary":"my ~/.claude directory","owner":"ymendel","repoUrl":"https://github.com/ymendel/dotclaude","license":"MIT","licenseTier":"permissive","trust":"community","stars":1,"match":null,"fewStars":true,"format":"skill","createdAt":"2026-04-08T12:58:02.000Z","addedAt":"2026-10-22T02:39:10.704Z","logoUrl":"https://avatars.githubusercontent.com/u/597?v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["writing"],"tags":[],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-26T10:12:23.116Z","sample":false,"url":"https://conduits.dev/tools/ymendel-dotclaude","language":"Shell","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# dotclaude\n\nYossef Mendelssohn's [Claude Code](https://claude.ai/code) config.\n\n## Consideration\n\nThis is personal config, and it's going to reflect my preferences, workflows, and the way I think about things. I'm not against others using any of it, but I'd encourage you to treat it as inspiration rather than installation.\n\n## What's Here\n\n- **CLAUDE.md** — the top-level instructions Claude always has in context: communication style, behavior, &c.\n- **[rules/](rules/README.md)** — more detailed rules, also always loaded: code style, development workflow, honesty, ADRs, self-improvement, RTK\n- **[skills/](skills/README.md)** — invokable skills that extend Claude's capabilities for specific tasks\n- **[agents/](agents/README.md)** — custom subagent configurations\n- **settings.json** — Claude Code settings\n- **hooks/** — the `PreToolUse` guards and command rewrites registered in `settings.json`, with checks over them in `hooks/test/` (`./hooks/test/run-checks.sh`)\n- **[scripts/](scripts/README.md)** — tooling for maintaining skills across this repo and a separate team skills repo\n- **[docs/adr/](docs/adr/)** — architecture decision records\n\n## Prerequisites\n\nGrouped by how routinely an absence bites, since that varies more than the flat list used to suggest. [Claude Code](https://claude.ai/code) itself goes without saying. The tiers are also the exit-code contract of the check at the end of this section.\n\n**Required — the absence breaks something broadly, and not always loudly:**\n\n- **[jq](https://jqlang.github.io/jq/)** — every `PreToolUse` hook parses its input with it. Without jq they all pass through, so the reflexive-cd guard, the shell-machinery guard and the uv-run guard are *off*, not merely quiet. Only `reflexive-cd-guard.sh` and the vendored `rtk-rewrite.sh` say anything about it; the rest exit in silence.\n- **[Python 3](https://www.python.org/)** — runs the session-handoff scripts, `scripts/rules-sections.py`, and the skill validators, and the `python`→`python3` rewrite hook targets it. Without it, handoff creation and loading fail. Loud, at least.\n\n**Load-bearing — the absence stops or slows ordinary work, loudly:**\n\n- **[RTK](https://github.com/rtk-ai/rtk)** — the hooks and rules assume it. Without it the rewrite hook passes commands through unchanged and warns, so nothing breaks — but a good share of the Bash allow rules are written in `rtk …` form and stop matching, which means approval prompts for commands you already granted. The lost token savings are the smaller half of this.\n- **[gh](https://cli.github.com/)** — the development-workflow rules reach for it constantly: PR and issue state, review comments, the checks before a merge is called ready. Without it that whole surface is unreachable from a session, and there is no in-session fallback. `git` itself is untouched, so pushing and pulling still work.\n- **[uv](https://github.com/astral-sh/uv)** — runs three things: the `ascii-diagram-validator` skill's script, which any session can invoke; the skill-architecture skill's validator and scaffolding scripts, which is what `uv-run-guard.sh` guards; and the Python test suite, which has no other working invocation because the system `python3` carries no pytest. Without it the first of those fails wherever the skill fires. See [ADR 0008](docs/adr/0008-testing-convention-for-hooks-and-scripts.md) for the tier.\n\n**Optional — the absence shows only on a path you may never take:**\n\n- **[trafilatura](https://github.com/adbar/trafilatura)** — extracts a web page's main content as markdown, the first choice for reading a page under the searching rules (`uv tool install trafilatura`). Without it those fetches fall back to `curl` for raw HTML or WebFetch for a summary, both of which still work — you lose a compact verbatim option, not a capability.\n\nTo verify these are on your `PATH`, run `./scripts/check-prerequisites.sh` — it reports each one with its tier, and exits 1 if anything in the required tier is missing. A missing load-bearing or optional prerequisite is reported and leaves the exit code at 0.\n\n## Installation\n\nThere's a Makefile to make this simple. Just clone and go.\n\n    $ git clone https://github.com/ymendel/dotclaude.git\n    $ cd dotclaude\n    $ make\n\nSee [the Makefile](Makefile) for the specific steps. In short, it symlinks the repo to `~/.claude`.\n\n## Appreciation\n\n- [softaworks/agent-toolkit](https://github.com/softaworks/agent-toolkit) — several agents and skills adapted from here, including session-handoff\n- [VoltAgent/awesome-claude-code-subagents](https://github.com/VoltAgent/awesome-claude-code-subagents) — source for many of the agents\n- [terrylica/cc-skills](https://github.com/terrylica/cc-skills) — ascii-diagram-validator and skill-architecture skills\n- [supabase/agent-skills](https://github.com/supabase/agent-skills) — supabase-postgres-best-practices skill\n- [mattpocock/skills](https://github.com/mattpocock/skills) — tdd and knowledge-grill skills\n- [alkofu/ai-tpk](https://github.com/alkofu/ai-tpk) — commit-message-guide skill\n- Flagrant — adr, adr-refine, purposeful-commits, and rails-test-discipline skills\n- [Chris Arcand — Purposeful Commits](https://chrisarcand.com/purposeful-commits/) — the method the purposeful-commits skill is built on\n- [fredrik-hansen](https://github.com/fredrik-hansen/fredrik-hansen/blob/main/HONESTY_RULES.md) — the honesty rules started here\n- Strunk & White's *The Elements of Style* — the writing skill is built around it\n\n## License\n\n[MIT](LICENSE). Skills adapted from upstream projects carry an `ATTRIBUTION.md` preserving the upstream notice. See each skill's directory.\n"}