- 📄 LICENSE
- 📄 README.md
- 📄 SKILL.md
AKTIVIERT SICH AUTOMATISCH bei vagen Auftraegen. LIEBER EINMAL ZU OFT NACHFRAGEN als falsch implementieren. Erkennungsmerkmale (EINES genuegt!): - Auftrag <25 Woerter - Keine konkreten Dateinamen/Pfade - Vage Verben: besser, optimieren, fixen, machen, aendern, verbessern, anpassen, erweitern, refactoren, aufraumen, ueberarbeiten - Unsichere Sprache: irgendwie, vielleicht, mal eben, schnell, einfach, bisschen, koennte, sollte - Fehlende Erfolgskriterien: Kein damit, sodass, weil, um zu - Relative Begriffe ohne Kontext: schneller, besser, schoener, einfacher Output ist STRUKTURIERTES JSON fuer prompt-architect Skill.
Практическая инженерия диффузионных моделей: архитектуры, обучение, инференс, оптимизация памяти. Использовать при любых задачах с диффузионными моделями: проектирование или модификация архитектуры (UNet/DiT/Flow/Flux), выбор и настройка schedulers/samplers, дообучение (LoRA/DreamBooth/full fine-tune), оптимизация памяти (AMP/checkpointing/ZeRO/FSDP/quantization), замена или fusion текст-энкодеров (CLIP/Qwen), работа с Diffusers, отладка диффузионных пайплайнов, оценка качества (FID/CLIPScore/LPIPS), latent diffusion, VAE, guidance/CFG, rectified flow, Stable Diffusion, SDXL, Flux. Также применять при вопросах про GPU-память при обучении генеративных моделей, text-to-image пайплайны, ControlNet, multi-encoder fusion, WebDataset. --- # Diffusion Engineering Skill ## Быстрая ориентация Три инженерных решения, которые больше всего влияют на качество/скорость/стоимость: 1. **Где идёт диффузия** → пиксели (дорого) или латентное пространство (LDM/SD-семейство — практично) 2. **Backbone денойзера** → UNet (классика, проще) или Transformer/DiT/Flow (масштабируется лучше) 3. **Управление сэмплингом** → scheduler, число шагов, guidance_scale — часто дают больше, чем правка сети --- ## Reference files — читать по задаче | Тема | Файл | Когда читать | |---|---|---| | Архитектуры и data flow | `references/architectures.md` | DDPM/SDE/LDM/DiT/Flux/VAE/SDXL, схема пайплайна | | Schedulers и guidance | `references/samplers.md` | DDIM/Euler/Heun/DPM-Solver/PNDM, CFG, prediction_type | | Обучение и дообучение | `references/training.md` | Loss/цели, LoRA/DreamBooth/full FT, гиперпараметры | | Память и распределённость | `references/memory.md` | AMP, checkpointing, ZeRO, FSDP, quantization, FP8 | | Текст-энкодеры и данные | `references/encoders-data.md` | CLIP/Qwen/multi-encoder, токенизация, data pipeline | | Оценка и траблшутинг | `references/eval-debug.md` | FID/CLIPScore/LPIPS, типовые поломки и фиксы, лицензии | --- ## Быстрый чеклист «я строю/модифицирую diffusion» - [ ] **Backbo
Workflow guide for querying Italian ISTAT statistical data via this MCP server. Use this skill whenever working with ISTAT data, SDMX dataflows, Italian statistics, regional/provincial data, unemployment, population, GDP, agriculture, or any other ISTAT dataset. Guides the discover -> constraints -> data workflow step by step.
Fetch and displays ABAP system information such as system ID, client, and user details.
- 📄 .gitignore
- 📄 clawhub.json
- 📄 config.json.template
使用 data.diemeng.chat 提供的接口查询股票日线、分钟线、财务指标等数据，支持 A 股等市场。
Turns prose, bullets, or structured notes into DiagramForge-compatible diagram source. Prefer Mermaid for graph-native diagrams and the Conceptual DSL for slide-native layouts such as matrix, cycle, funnel, chevrons, radial, pillars, and pyramid.
- 📁 references/
- 📁 scripts/
- 📄 SKILL.md
Build search applications and query log analytics data with OpenSearch. Use this skill when the user mentions OpenSearch, search app, index setup, search architecture, semantic search, vector search, hybrid search, BM25, dense vector, sparse vector, agentic search, RAG, embeddings, KNN, PDF ingestion, document processing, or any related search topic. Also use for log analytics and observability — when the user wants to set up log ingestion, query logs with PPL, analyze error patterns, set up index lifecycle policies, investigate traces, or check stack health. Activate even if the user says log analysis, Fluent Bit, Fluentd, Logstash, syslog, traceId, OpenTelemetry, or log analytics without mentioning OpenSearch.
Interact with data.gouv.fr APIs — Main API (datasets, orgs, users, resources, reuses, discussions), Metrics API (usage/stats by model), Tabular API (query CSV rows by resource ID). Use when working with data.gouv.fr data, catalog, or platform features.
- 📄 implementation.ts
- 📄 SKILL.md
Fetches ALL open issues from any GitHub repository using pagination and generates a comprehensive analysis including category breakdown, age distribution, stale issues (30+ days), top discussed issues, prioritization, and detailed recommendations for triage. Handles large repositories (5000+ issues) efficiently.
Dump PMM memory state as ASCII art in the terminal. Three depth levels: status (heatmap only), summary (heatmap + clusters + timeline), detailed (full ASCII). Runs as a subagent. Use when the user runs /pmm-dump or asks for a text-based memory overview. Trigger on: "pmm-dump", "/pmm-dump", "dump memory", "ascii memory", "text memory overview", "show memory heatmap", "memory dump", or any request for a text-based ASCII visualization of memory state. --- # PMM Dump Render PMM memory state as inline ASCII visualizations. Runs as a subagent to keep the main context clean. **Depth level:** $ARGUMENTS (default: `status` if empty or not provided) ## Invocation - `/pmm-dump` or `/pmm-dump status` — heatmap only (status level) - `/pmm-dump summary` — heatmap + cluster list + last 5 timeline entries - `/pmm-dump detailed` — full ASCII: graph map + heatmap + similarity matrix + clusters ## Behaviour Dispatch a `general-purpose` agent using the `Readonly Agent Model` from `memory/config.md` (default: `haiku`) with the prompt below. Replace `<level>` with the depth level (`status`, `summary`, or `detailed`). Replace `<project-root>` with the actual project root path. Output the agent's returned string verbatim — it contains the fully formatted ASCII visualization. ### Agent Prompt > Render PMM memory state as ASCII visualizations. This is a READ-ONLY task — do not edit any files. You may run git commands for timestamps. > > **Project root:** `<project-root>` > **Depth level:** `<level>` > > ### Depth Levels > > - `status` — Heatmap only > - `summary` — Heatmap + cluster list + last 5 timeline entries > - `detailed` — Full ASCII: graph map + heatmap + similarity matrix + clusters > > ### Visualization 1: Heatmap — File Activity (all levels) > > 1. Read `<project-root>/memory/config.md` to get the list of active files > 2. For each active file, run: `git log -1 --format="%ar|%at" -- memory/<filename>` > 3. Map the unix timestamp to a heat level: > - `████` = modified < 5 minute
Triage static analysis findings, assess merit, and accept noise or irrelevant items
Network reconnaissance and AI/ML service detection. Scan IP ranges with ping sweeps, port scanning, DNS resolution, and AI service probing across 45 detection signatures. Use when the user wants to discover hosts, open ports, or AI/ML services on a network.
