cc-best-config

Analyze structured data and produce evidence backed reports with Python. Use when the user asks to analyze CSV, Excel, SQL results, KPIs, business metrics, or other tabular data, including requests like "look at this data", "what trends do you see", "make a chart", "summarize these numbers", or "write a report from this spreadsheet". Also trigger for research synthesis from multiple sources and for database table analysis involving systems like ODPS, BigQuery, or MySQL. This skill covers data requirements, sufficiency checks, Python analysis, charts, and a polished Markdown report with limits and sources.

Set up an autonomous improvement loop for code, config, or content. The agent edits a target, runs an evaluation, and keeps iterating until a measurable metric improves or a stop condition is met. Use when the user wants repeated automatic attempts rather than one pass, especially for requests like "自动迭代优化", "keep trying until", "一轮一轮改到全部 pass", "auto-research", or "experiment loop". Trigger when the user provides both an editable target and a scriptable success metric such as test pass count, benchmark score, coverage, accuracy, bundle size, or validation loss.

Search, read, and browse academic papers via the Hugging Face CLI (hf papers). Use when the user wants to find papers, read a paper, check daily/trending papers, or get paper details. Triggers include: 搜论文, 找论文, 读论文, 看论文, paper search, read paper, daily papers, trending papers, arXiv, 论文检索, 论文阅读, what papers, latest papers, paper info, hf papers. Do not use for downloading model weights, datasets, or non-paper HF Hub operations.

AI 画图和图片生成技能。Use when the user wants illustrations, posters, avatars, style transfer, reference image generation, multi image style fusion, character consistency, or face stylization. Supports Gemini and OpenAI compatible image APIs, custom aspect ratios, high resolution output, and reference images. Do not use for image processing tasks such as compression, cropping, stitching, watermarking, format conversion, or code that manipulates images with PIL, OpenCV, or ffmpeg. Triggers include 画图, 生成图片, 画插画, 画海报, 风格转换, 参考图, 融合风格, 角色一致, 生成头像, generate image, illustration, style transfer, reference image, AI art, face editing, and image gen.

Delegate tasks to external CLI coding agents (Codex, OpenCode, QoderCLI) via Agent Bridge. Replaces cli-agents for external-agent work. Use when the user wants to send work to Codex, OpenCode, or QoderCLI for code review, adversarial review, explanation, implementation assistance, or general task delegation. Supports multi-agent execution with --agents for comparing results. Trigger on: 'use codex to review', '让 opencode 处理', 'delegate to qoder', 'external agent', 'agent review', 'review with codex and opencode', 'explain with opencode', 'let codex handle this', any mention of codex, opencode, qoder, or qodercli, or review/explain/delegation requests that explicitly involve external CLI agents.

Run a worker critic quality loop for non-trivial deliverables. One or more Worker agents produce drafts, then an independent Critic evaluates them against an explicit rubric and drives revision until the output passes. Use when quality is judged by criteria such as coverage, accuracy, depth, structure, or decision quality, especially for research, specs, RFCs, runbooks, analyses, and code with meaningful design choices. Trigger when the user asks for review, critique, a second opinion, a reviewer or critic agent, a quality gate, or high-confidence output. Also trigger proactively when the task has several quality dimensions, the result matters to stakeholders or integration, and missing gaps would cause expensive rework. Do not use for tiny edits, simple questions, or tasks optimized by scriptable numeric metrics; use auto-research for metric driven loops.

Generate Excalidraw diagrams from natural language descriptions. Use when asked to "create a diagram", "make a flowchart", "visualize a process", "draw a system architecture", "create a mind map", or "generate an Excalidraw file". Supports flowcharts, relationship diagrams, mind maps, and system architecture diagrams. Outputs .excalidraw JSON files that can be opened directly in Excalidraw.

将 Markdown 中的本地图片通过 PicList 上传到用户配置的图床，用在线 URL 替换本地路径。 依赖本地运行的 PicList App（HTTP API 127.0.0.1:36677），复用用户已有的图床配置。 使用场景：Markdown 含本地图片且需发送到外部平台（钉钉文档、邮件、分享报告等）时， 必须先用此 skill 将本地路径替换为可公网访问的链接。 触发词：本地图片上传、图片转链接、图床、piclist、markdown图片替换、 upload images、image hosting、钉钉文档图片。

Help users plan non-trivial engineering work before implementation. Use when the task involves migrations, refactors, architecture choices, project structure, risky multi-file changes, or feature strategy. Trigger on explicit requests such as "plan this", "break this down", "how should I approach", "design this feature", "review my approach", "帮我规划", or "怎么分步骤做", and also when the user presents a complex coding situation with real tradeoffs and needs safe sequencing. Do not use for simple scripts, isolated one file fixes, data analysis, or questions that only need a direct answer.

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.