数据与 AI技能

Diagnose whether an LLM classifier's validation-gate failure is GOLD-BOUND before spending on prompt revision or model changes. Use when: (1) a scoring pipeline over-predicts a label (precision low, recall high) and a prompt clarification is proposed to tighten it, (2) a pilot/validation gate fails and the fix candidates are prompt edits, (3) inter-rater agreement on the weak label was already low (κ < ~0.6). Core check: if gold POSITIVES share the exact feature the revision would exclude, no prompt can pass a gold-scored gate — recall craters while precision barely moves. Also documents the verified surgical-pilot design (single-section diff, tune/holdout split, pre-registered gate, perturbation check on untouched sections).

中文简介正在准备中。当前显示原文说明。

Build human adjudication / hand-labeling sheets from LLM-pipeline data without evidence truncation. Use when: (1) preparing a CSV/Excel sheet for a human to rule on cases an LLM classifier or rater panel judged, (2) a labeler reports "there is no information to label from" or cells look empty in Excel, (3) excerpt columns cluster at one exact length (e.g. all 1,500 chars — a hard truncation cap). Covers: full rating-basis recovery, Excel 32,767-char cell cap, multi-line CSV mangling, ruling dropdowns, companion text files.

中文简介正在准备中。当前显示原文说明。

Connect to and query WRDS (Wharton Research Data Services) from any research project. Use this skill whenever the user needs to download, query, or explore data from WRDS — including Compustat, CRSP, FactSet, I/B/E/S, or any other WRDS-hosted database. Also trigger when the user mentions WRDS tables, WRDS libraries, or wants to look up variable definitions or coverage in WRDS datasets. Do NOT trigger for general SQL or database questions unrelated to WRDS.

中文简介正在准备中。当前显示原文说明。

Use this skill whenever the user asks to run Stata commands, estimate econometric models, work with .dta files, run a .do file, generate Stata output, or do any statistical analysis where Stata is involved. Also trigger when the user mentions Stata variables, Stata syntax, or econometric tasks where Stata is the natural tool, including regressions, IV estimation, diff-in-diff, RDD, panel data, clustering, summary statistics, and margins. Stata runs through pystata on StataNow 19.5 BE; configure once with stata_setup, then drive everything with stata.run() and exchange data directly with pandas. Prefer this skill over subprocess calls or .do-file shelling for Stata work, including cases where the user does not say pystata.

中文简介正在准备中。当前显示原文说明。

Style every matplotlib figure like the Stata 18/19 default (stcolor) scheme — Arial embedded as TrueType, white background, recessive light-gray grid below the data, no top/right spines, unframed legends, and the validated blue/red/gray palette. Use whenever a task generates or restyles charts, plots, or figures for papers, reports, or slides, even if the user doesn't mention Stata — this is the house style for all publication figures. Also use when asked to make figures "look like Stata", match the stcolor scheme, or restyle existing matplotlib output.

中文简介正在准备中。当前显示原文说明。

Gate resumption of any multi-day LLM batch-scoring campaign that calls an unpinned model alias (deepseek-chat, gpt-*-latest, gemini-*-preview, any provider alias without a pinned version). Use when: (1) resuming a paused or credit-exhausted scoring run days after its last chunk, (2) topping up credits to finish a campaign, (3) extending a cached scoring pipeline with new items. Prevents silently splicing two model versions or serving revisions into one measure. Verified 2026-07-16: for $0.30 caught a serving-revision drift WITHIN DeepSeek v4-flash (same alias, same family, litigation scores systematically shifted across a 2-day gap) before an $83 resume spend.

中文简介正在准备中。当前显示原文说明。

Raise real concurrency in asyncio LLM batch scorers built on the OpenAI SDK (AsyncOpenAI, including OpenAI-compatible providers like DeepSeek). Use when: (1) raising an asyncio.Semaphore above ~100 produces no throughput gain, (2) a batch pipeline saturates near 100 in-flight requests despite a larger semaphore, (3) planning a high-concurrency campaign against a provider with no hard rate limit (DeepSeek v4-flash tolerates 2000+ in flight). Root cause: AsyncOpenAI's default httpx pool caps max_connections at 100, silently bottlenecking any larger semaphore — you must pass a custom http_client with httpx.Limits sized to the semaphore.

中文简介正在准备中。当前显示原文说明。

Diagnose and fix slow pyfixest regression GRIDS (many feols/fepois calls run sequentially) that stay slow despite demeaner_backend="cupy64" and an idle GPU. Use when: (1) a script looping dozens of pf.feols models on a 100k+ row panel takes ~1 min/model, (2) process inspection shows ~1-1.5 cores busy and nvidia-smi shows ~0% GPU utilization with a resident cupy context, (3) planning any worker prompt that will run a model grid (robustness variants x FE structures x domains). Root cause: per-model CPU-side single-threaded fixed costs (formulaic model-matrix build, interaction construction, singleton detection, cluster vcov) dominate wall time; GPU demeaning is a small slice. Fix: shard the model grid across OS processes and/or use pyfixest multiple-estimation syntax; mandate this IN THE WORKER PROMPT.

中文简介正在准备中。当前显示原文说明。

Judges whether a user's input suits the Vibe Creating style of video-prompt writing, and when it does, distills single-scene prompts, multi-shot descriptions, emotional imagery, or mixed input into prompts that are easier for a video model to generate from — while preserving any user-specified dialogue, voiceover, music, sound effects, and other hard constraints. Use when a user wants to turn an idea, story, feeling, or rough/over-specified prompt into a strong text-to-video prompt (Seedance, Sora, Kling, Veo, Runway, etc.), or asks to "rewrite", "improve", "clean up", or "vibe-ify" a video prompt. Do NOT use for long narrative films that need precise word-for-word dialogue sync, industrial shot lists meant to be executed verbatim, or functional/UI demos and step-by-step tutorials.

中文简介正在准备中。当前显示原文说明。

**WORKFLOW SKILL — MANDATORY for ANY hook work.** Create, configure, edit, debug, or review lifecycle hooks for VS Code Copilot (default) and Claude Code. USE FOR: criar/configurar/editar hook, hook do Copilot/Claude/agente, lifecycle hook, PreToolUse, PostToolUse, SessionStart, SubagentStop, Stop hook, UserPromptSubmit, PreCompact, bloquear comando, validar tool call, injetar contexto, guard, .agent.md frontmatter hooks, hooks.json, .github/hooks, .claude/settings.json, hook Node.js, cross-platform, hook não dispara/silencioso/ignorado, permissionDecision, hookSpecificOutput, additionalContext. **DEFAULT = VS Code GitHub Copilot.** Switch to Claude Code only when user explicitly says "Claude" OR workspace has `.claude/`. NEVER assume Claude. DO NOT USE FOR: general coding, creating agents (use agent-creator), creating skills (use skill-creator).

中文简介正在准备中。当前显示原文说明。

**CONVERSION SKILL** — Convert binary/Office documents to Markdown for LLM consumption. USE FOR: converting Word, Excel, PowerPoint, PDF, Visio, MS Project files to Markdown, batch document conversion, extracting text from non-readable formats. DO NOT USE FOR: Markdown formatting/linting, general file I/O, web scraping, image generation.

中文简介正在准备中。当前显示原文说明。

Enhance and improve existing AI prompts using professional techniques from a 2,500+ prompt database. Adds detail, fixes structure, optimizes for specific models, and applies proven patterns from top-performing prompts. Use when user says "enhance my prompt", "improve this prompt", "make this prompt better", "optimize prompt", "fix my prompt", or "upgrade this prompt".

中文简介正在准备中。当前显示原文说明。

Adapt and convert AI prompts between different models and platforms. Translates between Midjourney, Flux, Leonardo AI, DALL-E, Sora, Imagen, Stable Diffusion, Adobe Firefly, Ideogram, and more. Handles syntax differences, parameter mapping, and model-specific optimizations. Use when user says "adapt prompt", "convert prompt", "translate prompt", "port to flux", "midjourney to dall-e", or "change model".

中文简介正在准备中。当前显示原文说明。

Browse, filter, and explore a curated library of 2,500+ AI prompts organized by 19 categories, 17 models, and 4 output types. View stats, discover trending styles, and export prompt collections. Use when user says "prompt library", "browse prompts", "show categories", "list prompts", "prompt collection", "explore prompts", or "what prompts do we have".

中文简介正在准备中。当前显示原文说明。

Build custom AI prompts from scratch using guided workflows and patterns from a 2,500+ prompt database. Supports image, video, and text prompts for Midjourney, Flux, Leonardo AI, DALL-E, Sora, and 12 more models. Use when user says "build a prompt", "create a prompt", "write a prompt", "make me a prompt", "new prompt", or "prompt from scratch".

中文简介正在准备中。当前显示原文说明。

Measure writing-style statistics (sentence length, vocabulary, transitions, phrasing) from 5 to 10 posts you or your client wrote, and produce a voice profile for VOICE.md, blog-persona, and blog-write. Use for "learn my style", "build a writing baseline". It describes style; it does not detect or attribute authorship, and it is not for scoring a post (blog-analyze) or writing brand rules (blog-brand).

中文简介正在准备中。当前显示原文说明。

Research what people are actually saying about a topic in the last 30 days across Reddit, X / Twitter, YouTube, Hacker News, dev.to, Medium, and other public discourse platforms. API-free; uses WebSearch with platform-targeted site operators plus recency filters. Produces DISCOURSE.md (a structured brief) and JSON output the writer can consume. Complements blog-researcher (which focuses on authority sources) with a recency-and-engagement lens. Not for verifying statistics (blog-factcheck) or gathering authority sources. Use when user says "blog discourse", "discourse research", "what are people saying about", "research what people are saying", "voice of customer", "social listening", "30-day research", "trend research", "what's the discussion on", "real-time research", "practitioner discourse", "/blog discourse".

中文简介正在准备中。当前显示原文说明。

Internal capability, not user-invocable: the orchestrator calls it while running blog-write or blog-rewrite to turn research data into dark-mode compatible inline SVG charts (horizontal bar, grouped bar, donut, line, lollipop, area, radar) for HTML or MDX. Produces accessible markup (role=img, aria-labelledby, unique ids), source attribution and transparent backgrounds. Agents cannot call it.

中文简介正在准备中。当前显示原文说明。

FLOW framework integration for bloggers. Evidence-led content workflow using the Find, Optimize, Win loop with stage-specific AI prompts from the FLOW knowledge base (30 blog-applicable prompts, adapted from FLOW under CC BY 4.0). Use when user says "FLOW framework", "blog flow", "/blog flow", "evidence-led blogging", or "find optimize win", or asks for stage-specific blog prompts. Not for generic prompt writing.

中文简介正在准备中。当前显示原文说明。

Establish durable brand and voice context for cross-skill consumption. Generates BRAND.md (audience, positioning, do/don't editorial rules, taboo phrases, competitor differentiation) and VOICE.md (existing persona JSON re-expressed as readable prose), both written to the project root. When present, the blog orchestrator loads these files, fenced as untrusted data, for write, rewrite, brief, outline, calendar, and strategy workflows. Pairs with blog-persona, which manages the structured persona JSON. Use when the user says "blog brand", "create brand context", "BRAND.md", "VOICE.md", "establish editorial brand". Not for learning a voice from existing posts (blog-style) or managing persona JSON (blog-persona).

中文简介正在准备中。当前显示原文说明。

SERP-informed outline generation with H2/H3 heading hierarchy, competitive content gap analysis, section-by-section word count targets, chart and image placement markers, optional FAQ question planning, and internal linking zones. Skeleton only: structure, H2/H3 hierarchy, word counts, optional FAQ slots. Use blog-brief instead if you need full competitive analysis, statistics research, and image suggestions. Lighter than a full content brief, generates article skeleton and structure only, ready for /blog write to consume. Use when user says "outline", "blog outline", "content outline", "structure blog", "plan sections", "article skeleton", "heading structure", "SERP analysis", "competitive outline", "plan article".

中文简介正在准备中。当前显示原文说明。

Create a content brief for ONE planned post: target and secondary keywords, search intent, template choice, competitor gaps from the live SERP, sourced statistics to include, section plan, internal-link targets, and distribution ideas. Use when the user says "content brief", "blog brief", "SEO brief", "article brief". Not a skeleton only (blog-outline), a multi-post plan (blog-strategy, blog-calendar, blog-cluster), or the article itself (blog-write).

中文简介正在准备中。当前显示原文说明。

Professional finance research toolkit — backtesting (11 engines + benchmark comparison panel), factor analysis, Alpha Zoo (478 pre-built alphas across qlib158/alpha101/gtja191/academic/fundamental), options pricing, 90 finance skills, 30 multi-agent swarm teams, Trade Journal analyzer, and Shadow Account (extract → backtest → render) across 29 market-data sources (tushare, yfinance, okx, binance, akshare, baostock, tencent, mootdx, ccxt, futu, mt5, tickerall, local, eastmoney, sina, stooq, yahoo, pykrx, india_broker, qveris, longbridge, nobitex, wallex, plus optional-key finnhub/alphavantage/tiingo/fmp/gildata).

中文简介正在准备中。当前显示原文说明。

Run a Meta-Harness-style optimization loop NATIVELY — automatically search over the scaffolding around a FIXED base model (memory, retrieval, context construction, prompt templates, summarization, tool-selection logic) by proposing candidate variants, scoring each on a cheap deterministic eval, and keeping a Pareto frontier of quality vs cost — using native Agent / Workflow / loop tools instead of a standalone Python harness. Use this whenever the user wants to optimize, evolve, tune, distill, or search over a harness, scaffold, prompt system, memory or retrieval policy, context-assembly code, or summarizer while keeping the model fixed; whenever they mention Meta-Harness, harness optimization, scaffold evolution, automatic prompt/memory optimization, an evolutionary or Pareto search over candidate implementations, or "make the harness/agent better without retraining"; and whenever the gain must come from the code AROUND the model rather than the model weights. Reproduces the Meta-Harness paper's method natively, with no claude_wrapper.py and no metered solver API.

中文简介正在准备中。当前显示原文说明。