blender-kiln
這是一款 Claude Code 插件,能透過 Blender 將文字簡報或照片轉化為可投入生產的 3D 資產:透過程式碼重建、使用 CC0 掃描檔案進行紋理製作、根據真實物體進行色彩校正、優化為 GLB/FBX。
AI 翻譯
A small set of Claude Code skills: handoff, PR-comment handling, next-steps, README tidying, and TL;DR.
翻譯準備中
技能ShellGitHub @Slikon自行安裝營運者未確認
試用過了嗎? 點一下就好,不需要登入。
尚未偵測到更新。GitHub 發布版本後會顯示在這裡。
個人 GitHub 帳號下公開儲存庫的擁有者可以登入並驗證所有權,自行管理介紹,無需管理員核准。
登入並產生驗證檔案,在 30 分鐘內將其提交至此儲存庫的預設分支。其他分支或分叉儲存庫中的檔案無效。然後在下方檢查檔案,即可無需管理員核准直接管理專案。請保留檔案以維持管理權限。
如需登入,將前往登入頁面。登入後請回到此處,再次點擊所選的驗證按鈕。
安裝型專案無需部署網域也可驗證。個人儲存庫可透過 GitHub 登入驗證,組織儲存庫可使用上方的儲存庫驗證檔案。
製作者的介紹僅有上面的一句話。 更多詳情請閱讀儲存庫 README。 儲存庫 README ↗
正在載入完整 README。
檢視完整 README ↗A small set of agent skills for everyday engineering work. Each skill is a folder under skills/ with a single SKILL.md. The format is cross-agent — these install into Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, OpenCode, Zed, and 70+ other agents via the skills CLI.
| Skill | What it does |
|---|---|
teach-codebase | Build a mental model of an unfamiliar repo: architecture, subsystems, and where to look when changing something. Saved at .teach/codebase.md. |
teach-pr | Explain what a PR or branch actually does before you review it — mental model, traced execution path, what to question. Saved at .teach/. |
create-pr | Open or update a PR with a concise, human-first overview at the top, plus explicit notes and screenshots when UI or image size/quality changes. |
handoff | Compact the conversation into a handoff doc for the next agent. Saved at .handoffs/ so it survives a restart. |
respond-to-pr-comments | Work through a PR's review comments: fix the valid ones, reply to the rest explaining why not. Takes a PR url or number. |
next-steps | Read the latest handoff and lay out the next steps. |
readme-check | Make sure the project has a README that's short, complete, and free of fluff. |
tldr | Compress a wall of text into a tight summary. |
bro | Restate the last message in plain human language, with no jargon. |
teach-pr reads .teach/codebase.md when it exists and explains the change as a delta against it, so running teach-codebase once per repo makes every later teach-pr sharper. Both record the commit they were generated at, so a re-run updates what actually moved instead of regenerating from scratch. They're working notes, not docs — worth adding .teach/ to your .gitignore.
Needs Node (for npx) and git. respond-to-pr-comments also needs the gh CLI; teach-pr uses gh when a PR exists but falls back to a plain branch diff.
skills CLI (recommended, any agent)# Interactive — pick which skills and which agents
npx skills add Slikon/agent-skills
# Everything, globally, into Claude Code, no prompts
npx skills add Slikon/agent-skills -g -a claude-code -y
# Other agents, or several at once
npx skills add Slikon/agent-skills -a cursor -a codex
# Just one skill
npx skills add Slikon/agent-skills --skill tldr
Update later with npx skills update.
Clone and run the installer, which symlinks each skill into ~/.claude/skills/:
git clone https://github.com/Slikon/agent-skills.git
cd agent-skills
./install.sh # into ~/.claude/skills
CLAUDE_SKILLS_DIR=/path/to/project/.claude/skills ./install.sh # into one project
Restart your agent afterward so it picks up the new skills.
Type any skill by name:
/teach-codebase # or: /teach-codebase auth
/teach-pr # or: /teach-pr https://github.com/you/repo/pull/42
/create-pr
/handoff wiring up the payments webhook
/respond-to-pr-comments https://github.com/you/repo/pull/42
/next-steps
/readme-check
/tldr
/bro
Every skill except handoff and bro also carries a description, so the agent can fire it on its own from natural phrasing — "that's too long, give me the short version" fires tldr, "what's next?" fires next-steps. Those two are typed-only by design.