asgard-ai-platform/skills

ARIMA Time Series Model builds ARIMA or SARIMA forecasts from univariate historical data. Use it to test stationarity, choose parameters, and produce predictions with confidence intervals.

Ad Bidding Strategies helps choose and tune manual CPC, Target CPA, Target ROAS, and other bidding approaches. Use it when selecting a strategy, setting automated bids, or checking whether conversion data is sufficient.

Ad Budget Allocation optimizes spend across multiple campaigns using marginal returns analysis. Use it to split a fixed ad budget, rebalance after performance changes, or maximize ROAS under budget constraints.

Blockchain Fundamentals explains distributed ledger architecture, consensus, and block structure, and helps decide when blockchain is appropriate. It is used when evaluating a blockchain use case or choosing a permissioned or public design.

Bm25 Ranking implements the BM25 text ranking function for e-commerce product search. Use it to score and rank listings by keyword relevance, especially when replacing basic TF-IDF or tuning search results.

CTR Prediction Model estimates the probability that an ad will be clicked from user, query, ad, and position features. Use it for ad ranking, bid optimization, or evaluating creative performance.

Compensation Benchmarking compares internal pay to market survey data using compa-ratios and percentile positioning. Use it to assess market competitiveness, build salary bands, and check pay equity.

Cpk Process Capability calculates Cp and Cpk to check whether a stable process can meet specification limits. Use it to compare process capability, estimate defect rates, and assess centering against target values.

Design Of Experiments designs and analyzes factorial experiments to find significant process factors and interactions. Use it to screen variables, optimize settings, or confirm which inputs affect quality or yield.

E-Commerce Product Ranking combines relevance, CTR, conversion, ratings, and business metrics to sort products. Use it for product search or browse ranking when pure text relevance is not enough.

E-Commerce Search Relevance diagnoses and improves the full e-commerce search pipeline from query understanding to result presentation. Use it to reduce zero-result searches, add spell correction or synonyms, and troubleshoot relevance issues.

Employee Turnover Prediction builds models that estimate the probability an employee will leave and surfaces likely retention drivers. Use it for attrition prediction, flight-risk scoring, and prioritizing HR interventions.

Ensemble Forecasting combines predictions from multiple models to improve forecast reliability and reduce error. Use it when several ARIMA, Prophet, ETS, or ML forecasts are available and you need a robust combined prediction.

Exponential Smoothing applies SES, Holt, or Holt-Winters methods to forecast time series with level, trend, and seasonality. Use it for simple short-horizon forecasts and lightweight smoothing when a full model is unnecessary.

FMEA systematically identifies failure modes, their effects, causes, and controls to prioritize risks with RPN or Action Priority. Use it when assessing product or process risks, building a risk register, or planning corrective actions.

Gale-Shapley Stable Matching implements deferred acceptance for two-sided preference matching. Use it for candidate-to-job, student-to-school, or other stable assignment problems with ranked preferences.

Generalized Second Price Auction computes ad slot assignments and cost-per-click from bids and quality scores. Use it to analyze search ad auctions, ad rank, and bidding dynamics.

Influence Maximization selects k seed nodes in a network to maximize expected spread under Independent Cascade or Linear Threshold models. Use it to choose seeds for viral campaigns, maximize reach under a budget, or compare seeding strategies.

LDA Topic Modeling discovers latent topics in document collections and assigns documents to themes. Use it when you need exploratory topic extraction, document clustering by theme, or a summary of the main topics in a corpus.

Louvain Community Detection finds densely connected groups in weighted graphs by optimizing modularity. Use it to segment social, organizational, or customer networks and to analyze network structure.

Network Centrality calculates degree, betweenness, closeness, and eigenvector scores for nodes in a graph. Use it to identify key influencers, bridge nodes, and critical infrastructure in network data.

Prophet Forecasting builds business time series forecasts with trend, seasonality, holidays, and changepoints. Use it for daily or weekly metrics when you need a quick forecast with built-in handling for missing data and validation.

Smart Contracts designs and implements on-chain agreements, token logic, and other immutable blockchain programs. Use it when you need smart contract architecture, security review, or test coverage for Solidity or similar contracts.

Statistical Process Control implements control charts to monitor process stability over time and flag special-cause variation. Use it to set control limits, detect shifts, and tell whether a process is in control.

VCG Mechanism allocates ad slots to maximize welfare and computes externality-based payments for truthful bidding. Use it to design incentive-compatible auctions or compare VCG with GSP revenue tradeoffs.