AI & ML

moshe-ship Skill

مساعد صوتي عربي — حوّل الصوت لنص، عالجه، وردّ بصوت عربي. خط أنابيب كامل: Whisper → LLM → TTS. استخدم عندما يريد المستخدم تفاعل صوتي كامل.

moshe-ship Skill

ToolProof integration — verify agent tool calls, generate trust reports, analyze cost patterns

hibra999 Bundle

Use PyOD for anomaly/outlier detection after validating prepared tabular or time-indexed feature matrices, including classic fit/predict detectors, time-series detectors, ADEngine selection, contamination thresholds, anomaly scores, labels, probabilities, evaluation metrics, plotting, model persistence, and anti-leakage safeguards.

hibra999 Bundle

Use Facebook/Meta Kats for forecasting with TimeSeriesData, classical models, Prophet, LSTM, LightGBM MLAR, global models, ensembles, temporal hierarchical reconciliation, hyperparameter/meta-learning helpers, prediction intervals where documented, plotting, backtesting, and leakage-safe temporal validation. Trigger when an agent needs to model prepared time-series data with Kats after applying forecasting-data-prep for frequency, horizon, splits, covariates, and anti-leakage checks, especially when working with legacy Kats 0.2.0 projects.

hibra999 Bundle

Use Luminaire for time-series anomaly and outlier detection after validating ordered univariate data, including DataExploration profiling, HyperparameterOptimization, LADStructuralModel, LADFilteringModel, WindowDensityModel streaming/window detection, anomaly probabilities, confidence intervals, model freshness, and anti-leakage safeguards.

hibra999 Bundle

Use Unit8 Darts for forecasting with TimeSeries, local/statistical models, global regression and deep learning models, foundation models, covariates, static covariates, probabilistic forecasts, conformal prediction, multiple-series training, backtesting, residual diagnostics, plotting, and leakage-safe temporal validation. Trigger when an agent needs to model prepared time-series data with Darts after applying forecasting-data-prep for frequency, horizon, covariate availability, splits, and anti-leakage checks.

hibra999 Bundle

Use sktime for forecasting with its unified forecaster API, ForecastingHorizon, temporal splitters, pipelines, tuning, reductions to regression, statistical/deep/foundation/wrapped forecasters, exogenous variables, probabilistic forecasts, panel/global/hierarchical data, and leakage-safe backtesting. Trigger when an agent needs to model prepared time-series data with sktime after applying forecasting-data-prep for frequency, horizon, splits, covariates, and anti-leakage checks.

hibra999 Bundle

Use Salesforce Merlion for forecasting workflows after forecasting-data-prep, including TimeSeries/UnivariateTimeSeries data, DefaultForecaster, Arima, Sarima, ETS, Prophet, MSES, VectorAR, tree forecasters, DeepAR/Autoformer/ETSformer/Informer/Transformer, AutoETS/AutoProphet/AutoSarima, ForecasterEnsemble, exogenous regressors, uncertainty/error bars, ForecastEvaluator live-deployment validation, forecast metrics, plotting, and anti-leakage safeguards.

hibra999 Bundle

Use Prophet for univariate forecasting with trend, seasonality, holidays, extra regressors, uncertainty intervals, temporal cross-validation, and diagnostic plots. Trigger this skill when an agent needs to model a prepared time-series dataset with the official Python or R Prophet library, after first applying forecasting-data-prep to validate frequency, horizon, covariate availability, temporal splits, and anti-leakage safeguards.

hibra999 Bundle

Use sktime for time-series classification after ts-classification-data-prep, including Panel data formats such as numpy3D, pd-multiindex, df-list, nested_univ, equal/unequal length handling, univariate and multivariate samples, distance/kernel, dictionary, interval, feature, shapelet, ROCKET, hybrid, ensemble, early, deep learning and foundation classifiers, ClassifierPipeline, SklearnClassifierPipeline, predict/predict_proba, stratified cross-validation, model selection, padding/truncation, and anti-leakage safeguards.

hibra999 Bundle

Use Statsmodels for classical and econometric forecasting with AR/ARIMA/SARIMAX, exponential smoothing/ETS, state space models, VAR/SVAR/VECM/VARMAX, dynamic factor models, ARDL/UECM, ThetaModel, STLForecast, exogenous regressors, prediction intervals, temporal validation, and residual diagnostics. Trigger when an agent needs to model prepared time-series data with official statsmodels APIs after applying forecasting-data-prep for frequency, horizon, splits, covariates, and anti-leakage checks.

hibra999 Bundle

Prepare, validate, split, and document time-series classification datasets, then write problem-summary.txt for skill routing. Use this skill whenever an agent must define the classification problem, samples, labels, tensor or panel shape, channel order, padding or truncation policy, split IDs, class balance, group or time leakage risks, and preprocessing fit boundaries before classifier selection.

hibra999 Bundle

Use THUML Time-Series-Library (TSLib) for deep-learning time-series forecasting experiments with run.py, long_term_forecast, short_term_forecast, zero_shot_forecast, ETT/custom/M4 loaders, M/S/MS feature modes, seq_len/label_len/pred_len horizons, TimeXer exogenous workflows, deterministic metrics, temporal splits, and leakage-safe validation. Trigger when an agent needs to train, evaluate, or adapt a prepared forecasting dataset with TSLib after applying forecasting-data-prep.

hibra999 Bundle

Use THUML Time-Series-Library/TSLib for deep learning time-series classification after ts-classification-data-prep, including UEA .ts TRAIN/TEST files, UEAloader, variable-length padding masks, multivariate channels, task_name=classification, CrossEntropyLoss training, softmax probabilities, accuracy evaluation, official classification scripts, supported model names such as TimesNet, Transformer, Autoformer, PatchTST, iTransformer, MambaSingleLayer/MambaSL, TimeMixer, and strict anti-leakage preprocessing.

dubsopenhub Skill

🐝 Stampede Swarm — Hierarchical Agent Fleet

dubsopenhub Skill

🐝 Hive1K — recursive multi-model swarm orchestrator. Launches 250-1,000+ AI agents across 16 models with Division Commanders, hierarchical fan-out, Shadow Score Spec L2 conformance, and quality-gated synthesis. Say "hive1k" or "swarm1k" to start.

dubsopenhub Skill

Cross-terminal multi-agent orchestration. Splits complex tasks into parallel work units dispatched to independent Copilot CLI workers via tmux panes with filesystem IPC, atomic operations, dead worker recovery, and conflict-aware synthesis.

dubsopenhub Skill

🧬 Evo Forge — breeds better agents through evolutionary selection. Mutates agent prompts, tournaments them, selects survivors, tracks lineage. Say "evo forge" to evolve an agent, "evo forge status" to see the leaderboard.

dubsopenhub Skill

Agent Pulse — Copilot CLI Skill

dubsopenhub Bundle

Interactive training for the GitHub Copilot CLI. Guided lessons, quizzes, scenario challenges, and a full reference covering slash commands, shortcuts, modes, agents, skills, MCP, and configuration. Say "cliexpert" to start.

dubsopenhub Bundle

🎓 First Agent — live training skill that guides non-developers from zero to building their first AI agent in three sessions. Run It → Remix It → Build It. Designed for instructor-led sessions. Say "first-agent" to start.

dubsopenhub Skill

🔐 Sudo Summit — the nightly Agent Roundtable for your GitHub Copilot fleet. Agents request powers, grant superpowers to each other, and ship upgrades with one-tap permission (sudo). Runs a bounded, safe Town Hall, files asks, and — on your approval — applies granted upgrades to real agent files. Say "sudo summit" to start.

dubsopenhub Bundle

Agent Company — Startup in a Box

dubsopenhub Skill

Use when the user asks to "fetch context from Slack", "read this Slack thread", "get requirements from Slack", "extract Slack conversation", or provides a Slack permalink. Uses `gh slack read` — no background scraping, no MCP server. Always warn about security and request confirmation before using Slack data.

dubsopenhub Skill

🐝 Swarm Command — multi-model consensus swarm orchestrator. Launches 50-250+ AI agents across 15 models with hierarchical fan-out, cross-family review, Shadow Score Spec L2 conformance, and quality-gated synthesis. Say "swarm command", "swarmcommand", or "swarm250" to start.

dubsopenhub Skill

Multi-agent fleet conductor with real-time TUI observability. Launches multiple Terminal Stampede commander groups, keeps them coordinated through live collaboration ledgers, and applies sealed Shadow Score evaluation. Say "agent conductor" to start.

dubsopenhub Skill

Multi-agent fleet conductor for the GitHub Copilot CLI. Launches exactly five visible Terminal Stampede commander groups, coordinates them through live collaboration ledgers, watches them with Agent Pulse, and applies sealed Shadow Score evaluation. Say "agent orchestra" to start.

dubsopenhub Skill

Context Compass — your guide to bringing GitHub, Slack, and WorkIQ context into Copilot CLI. Runs beginner onboarding, daily briefings, safety checks, diagnostics, and role-based prompt recipes. Say "context compass" to start.

dubsopenhub Skill

Fleet Scorecard — turns any multi-agent CLI run into a clear outcome decision. Scores what changed, what won, what failed, and whether to run it again. Say "fleet scorecard" to start.

dubsopenhub Skill

🏟️ Havoc Hackathon — a multi-model orchestration skill that turns your terminal into a competitive arena. Dispatches the current selectable GitHub Copilot model roster in tournament elimination heats, scores them with sealed judge panels and Shadow Spec hidden quality gates, evolves the best ideas between rounds via Convergence Broadcasts, and synthesizes the final output from collective intelligence. Say "run hackathon" to start. Say "run kiloagent" for 1,000-agent deep mode. Say "run hackathon metaswarm" for commander contestants backed by full SS-250 swarms.

dubsopenhub Skill

✨ First Light — a warm, friendly guide that helps anyone build their first AI agent in about 10 minutes. No coding experience needed. Walks people through picking an agent, teaching it their voice, trying it out, and saving it — then reveals they just did what developers do every day.

x-cmd Bundle

任意 markdown 文章的 YAML front matter（YFM）。x-cmd 的默认有三个：`tags`、`description`、`memo`。默认借鉴 Agent Skills frontmatter 形态 —— 见 <https://agentskills.io/specification>。yfm 模块是个 helper：`x yfm ls / init / lint`。

x-cmd Bundle

agent 克隆仓库位置与 worktree 布局的约定 + 工具。默认 bare repo 在 `~/.x-repo/.bare/<provider>/<owner>/<name>.git`，linked working tree 在 `~/.x-repo/<provider>/<owner>/<name>[/@<wtname>]` 下。通过 `x repo ls` 发现。worktree 管理通过 `x repo wt`。适用于 "where to clone"、"git clone path"、"workspace layout"、"shared repos between agents"、"worktree path"。

x-cmd Bundle

为 AI agent 设计的加权维度打分框架 —— "Don't guess. Score." 核心理念：先理解需求、先草稿再迭代，agent 是分析师。通过 `x score compute` 输出排序后的 TSV。

x-cmd Bundle

x-cmd 的 prompt 工程约定 —— 通过模板变量复用、约定结构、安全强制模式。

x-cmd Bundle

Lovable MCP 工具用法 —— 部署项目、向 agent 发消息、用 agent-browser 测试。通过委托本地 repo 来减少 credits。