AI & ML

gptomics Bundle

Estimate divergence times under molecular-clock models with BEAST2, MCMCTree/PAML, TreePL, and LSD2, framing a date as a product of the calibration prior and the clock model far more than of the sequence data. Covers why branch length = rate x time is nonidentifiable so only calibrations convert relative rate-time into absolute age; why the effective (marginal) prior on a calibrated node differs from the density specified, mandating a sample-from-prior run; the fossil-as-minimum rule, soft bounds, tip-dating, and the fossilized birth-death process; the temporal-signal check (TempEst root-to-tip regression + date-randomization) required before dating viruses or ancient DNA; and clock-model choice via the coefficient of variation. Use when dating nodes, calibrating with fossils or sampling dates, choosing a clock or dating engine, or routing topology to modern-tree-inference, posteriors to bayesian-inference, and rooting to tree-manipulation.

gptomics Bundle

Edit phylogenetic tree structure with Biopython Bio.Phylo, and treat rooting as a separate statistical inference rather than a display choice. Covers why most inference returns an unrooted tree so placing the root creates every ancestor/descendant and basal claim; why a distant or lonely outgroup misroots inside the ingroup via long-branch attraction; the outgroup/midpoint/MAD/MinVar/non-reversible-likelihood rooting tradeoffs; why pruning must suppress degree-2 nodes and sum their branch lengths or all patristic distances silently corrupt; and why collapsing by support makes SOFT (uncertainty) polytomies, not HARD (radiation) ones. Use when rooting, re-rooting, pruning or subsetting taxa, extracting a clade or induced subtree, collapsing low-support branches, resolving polytomies, or ladderizing. Routes clock-based rooting to divergence-dating, inference to modern-tree-inference, and reading/plotting to tree-io and tree-visualization.

intent-solutions-io Bundle

Generate production-ready forecasting experiments with StatsForecast and TimeGPT. Use when setting up model benchmarking or cross-validation. Trigger with 'scaffold experiment' or 'compare models'.

intent-solutions-io Bundle

Fine-tunes TimeGPT on custom datasets to improve forecasting accuracy. Use when TimeGPT's zero-shot performance is insufficient or domain-specific accuracy is needed. Trigger with "finetune TimeGPT", "train TimeGPT", "adapt TimeGPT".

intent-solutions-io Bundle

Analyze Nixtla baseline forecasting results (sMAPE/MASE on M4 or other benchmark datasets). Use when the user asks about baseline performance, model comparisons, or metric interpretation for Nixtla time-series experiments. Trigger with "baseline review", "interpret sMAPE/MASE", or "compare AutoETS vs AutoTheta".

intent-solutions-io Bundle

Generate benchmarking pipelines to compare forecasting models and summarize accuracy/speed trade-offs. Use when evaluating TimeGPT vs StatsForecast/MLForecast/NeuralForecast on a dataset. Trigger with "benchmark models", "compare TimeGPT vs StatsForecast", or "model selection".

intent-solutions-io Bundle

Performs rigorous time series cross-validation using expanding and sliding windows. Use when needing to evaluate the performance of time series models on unseen data. Trigger with "cross validate time series", "evaluate forecasting model", "time series backtesting".

oimiragieo Skill

Agent frontmatter enhancements — disallowedTools, mcpServers scoping, fork_eligible field

gptomics Bundle

Aligns sample metadata with count matrices and constructs design matrices for downstream DE, handling the alphabetical-reference-level trap (relevel BEFORE DESeq), LRT reduced-model rules, the interaction-term resultsNames trap, continuous-covariate scaling and splines, repeated measures via duplicateCorrelation or dream, high-cardinality categorical pseudo-singular designs, sample swap detection via XIST/RPS4Y1 expression and somalier/NGSCheckMate genotypes, SABV (sex-as-biological-variable) mandate, Simpson's-paradox collapsing of technical replicates, and the `~ 0 + group` parameterization for clean contrasts. Use when building a design matrix, troubleshooting reversed fold-change direction, encoding paired or repeated-measures designs, detecting sample swaps, deciding sex-as-covariate, or aggregating technical replicates.

intent-solutions-io Bundle

Automatically selects the best forecasting model between StatsForecast and TimeGPT based on time series data characteristics. Use when unsure which model performs best. Trigger with "auto-select model", "choose best model", "model selection".

sboghossian-mini-claude-for-legal Skill

Use when designing the model-selection and fallback logic for a legal AI product — defining which model to use for which skill tier, how to cascade to a cheaper or faster model when the primary model is unavailable or over budget, and how to handle failures gracefully without exposing errors to legal practitioners. Engineering skill with direct impact on availability SLOs and cost management.

oimiragieo Bundle

Deeply analyzes Agent Studio framework structural health: catching phantom require() references, wrong module depth paths, missing skill/agent dependencies, bloated configurations, archived references in active code, stale catalog counts, and empty tool/skill directories.

intent-solutions-io Bundle

This skill enables Claude to split datasets into training, validation, and testing sets. It is useful when preparing data for machine learning model development. Use this skill when the user requests to split a dataset, create train-test splits, or needs data partitioning for model training. The skill is triggered by terms like "split dataset," "train-test split," "validation set," or "data partitioning."

intent-solutions-io Bundle

This skill trains machine learning models using automated workflows. It analyzes datasets, selects appropriate model types (classification, regression, etc.), configures training parameters, trains the model with cross-validation, generates performance metrics, and saves the trained model artifact. Use this skill when the user requests to "train" a model, needs to evaluate a dataset for machine learning purposes, or wants to optimize model performance. The skill supports common frameworks like scikit-learn.

gptomics Bundle

Basecalls raw Oxford Nanopore signal (POD5/FAST5) into reads with Dorado, choosing the chemistry-matched model and accuracy tier (fast/hac/sup), requesting modified bases (5mCG_5hmCG, 6mA, m6A) at basecall time, and handling duplex, demultiplexing, trimming, and HERRO read correction. Covers why the model+version is an irreversible analysis decision, why methylation cannot be recovered later, and why downstream polish/variant models must match the basecaller. Use when converting POD5/FAST5 to reads, picking a Dorado model for R9/R10 or RNA004, enabling methylation calling, basecalling duplex, demultiplexing barcoded runs, or correcting reads for assembly.

gptomics Bundle

Frames Bayesian phylogenetics as approximating a posterior distribution over trees conditioned on data AND priors via an MCMC that must be proven to have converged, using MrBayes, BEAST2, RevBayes, and PhyloBayes-MPI. Covers why convergence (ESS, PSRF, ASDSF, topology vs scalar) is the load-bearing claim, why posterior probabilities are systematically higher than bootstrap and overconfident under model misspecification, why the default branch-length prior inflates tree length, why the harmonic-mean estimator must never select models (use stepping-stone), and when site-heterogeneous CAT-GTR is required at depth. Use when needing posterior clade support, model averaging, marginal-likelihood model comparison, or CAT models for deep phylogeny. Routes topology-only ML to modern-tree-inference, divergence times to divergence-dating, and tree summarization to tree-io.

gptomics Bundle

Infers developmental trajectories, pseudotime, RNA velocity, and directed fate probabilities from single-cell data using PAGA, Slingshot, Monocle3, DPT, Palantir, scVelo, and CellRank 2. Use when ordering cells along a differentiation continuum, choosing a trajectory method by topology, rooting pseudotime, estimating RNA velocity direction, computing fate probabilities near a bifurcation, or judging whether an inferred trajectory is real.

gptomics Bundle

Integrates single-cell paired TCR/BCR (10x VDJ, AIRR, dandelion, BD Rhapsody) with gene expression in an AnnData/MuData object using scirpy - chain-pairing QC, clonotype definition, clonal expansion, diversity, repertoire overlap, V(D)J usage, and VDJdb specificity. Operates on the awkward-array AIRR model (adata.obsm['airr'], accessed via get.airr after pp.index_chains), not legacy per-chain obs columns. Use when deciding clonotype definition for TCR (exact CDR3-nt identity via define_clonotypes) versus BCR (nucleotide distance clustering via define_clonotype_clusters with normalized_hamming plus same_v_gene/same_j_gene, because somatic hypermutation shatters identity clonotypes); tuning receptor_arms (all vs any), dual_ir, and within_group; filtering chain_qc categories (multichain doublets, orphan dropout, extra-VJ dual-TCR) without biasing clonal-expansion and diversity estimates; and overlaying clonality onto the transcriptomic UMAP.

sboghossian-mini-claude-for-legal Skill

Use when an enterprise prospect or IT administrator asks about deploying the platform at scale — tenant isolation, SSO, audit logs, custom data residency, SLA, implementation timeline, and dedicated support. This is a platform documentation skill covering the enterprise deployment model, implementation phases, security architecture, and customization options for law firm and corporate legal department deployments.

intent-solutions-io Skill

This skill enables Claude to design NoSQL data models. It activates when the user requests assistance with NoSQL database design, including schema creation, data modeling for MongoDB or DynamoDB, or defining document structures. Use this skill when the user mentions "NoSQL data model", "design MongoDB schema", "create DynamoDB table", or similar phrases related to NoSQL database architecture. It assists in understanding NoSQL modeling principles like embedding vs. referencing, access pattern optimization, and sharding key selection.

intent-solutions-io Skill

This skill enables Claude to generate ORM models and database schemas. It is triggered when the user requests the creation of ORM models, database schemas, or wishes to generate code for interacting with databases. The skill supports various ORMs including TypeORM, Prisma, Sequelize, SQLAlchemy, Django ORM, Entity Framework, and Hibernate. Use this skill when the user mentions terms like "ORM model", "database schema", "generate entities", "create migrations", or specifies a particular ORM framework like "TypeORM entities" or "SQLAlchemy models". It facilitates both database-to-code and code-to-database schema generation.

intent-solutions-io Skill

This skill enables Claude to optimize machine learning model hyperparameters using grid search, random search, or Bayesian optimization. It is used when the user requests hyperparameter tuning, model optimization, or improvement of model performance. The skill analyzes the current context, generates code for the specified search strategy, handles data validation and errors, and provides performance metrics. Trigger terms include "tune hyperparameters," "optimize model," "grid search," "random search," and "Bayesian optimization."

gptomics Bundle

Predicts ADMET properties using ADMETlab 3.0 (119 platform features, including 77 prediction models with modeled-endpoint uncertainty), ADMET-AI, DeepChem MolNet, and chemprop D-MPNN with explicit handling of OECD QSAR principles, applicability domain assessment, calibration, hERG/CYP/AMES endpoints, and PAINS / Lipinski / Ro5 / Veber / BBB druglikeness filters. Use when filtering compounds for drug-likeness, prioritizing leads by predicted safety, or building an in-house ADMET QSAR model.

sboghossian-mini-claude-for-legal Skill

Use when discussing the legal talent + AI services market, the Axiom-Harvey partnership as a model for AI-augmented staffing, or distribution of legal AI tools through flexible-workforce and lawyers-on-demand networks. Covers the 2024 Axiom × Harvey deal, what it means for the ALSP sector, implications for BigLaw staffing economics, and the emerging "AI-equipped lawyer" service model relevant for MENA legal talent markets.

sboghossian-mini-claude-for-legal Skill

Use before delivering any substantive legal answer to score the model's own confidence in the proposed response. Produces a 0.0–1.0 confidence score and routes the request to one of four handling modes — proceed, hedge, cite-or-bust, or escalate — based on jurisdiction coverage, recency requirements, specificity of the claim, and stakes. The anti-hallucination gate for all legal AI output.

intent-solutions-io Skill

This skill optimizes prompts for Large Language Models (LLMs) to reduce token usage, lower costs, and improve performance. It analyzes the prompt, identifies areas for simplification and redundancy removal, and rewrites the prompt to be more concise and effective. It is used when the user wants to reduce LLM costs, improve response speed, or enhance the quality of LLM outputs by optimizing the prompt. Trigger terms include "optimize prompt", "reduce LLM cost", "improve prompt performance", "rewrite prompt", "prompt optimization".

intent-solutions-io Skill

This skill allows Claude to evaluate machine learning models using a comprehensive suite of metrics. It should be used when the user requests model performance analysis, validation, or testing. Claude can use this skill to assess model accuracy, precision, recall, F1-score, and other relevant metrics. Trigger this skill when the user mentions "evaluate model", "model performance", "testing metrics", "validation results", or requests a comprehensive "model evaluation".

intent-solutions-io Skill

This skill allows Claude to construct and configure neural network architectures using the neural-network-builder plugin. It should be used when the user requests the creation of a new neural network, modification of an existing one, or assistance with defining the layers, parameters, and training process. The skill is triggered by requests involving terms like "build a neural network," "define network architecture," "configure layers," or specific mentions of neural network types (e.g., "CNN," "RNN," "transformer").

gptomics Bundle

Bead-based signal normalization and cross-batch harmonization for CyTOF and high-parameter cytometry - EQ four-element bead normalization of instrument sensitivity drift (CATALYST normCytof, premessa), and reference-anchor cross-batch normalization (CytoNorm, per-cluster quantile splines). Covers the distinction between within-run drift correction and between-batch correction, the mandatory anchor/reference sample, why normalization is per-cluster with many quantiles, and the over-correction risk. Use when correcting CyTOF signal drift, harmonizing multi-batch or multi-site studies, or deciding whether to normalize data versus model batch in the design.

gptomics Bundle

Parses, queries, converts, and extracts from GTF and GFF3 gene-model annotation files - walking the gene/transcript/exon/CDS hierarchy with gffutils (queryable SQLite DB), converting formats and extracting transcript/CDS/protein FASTA with gffread, slurping to dataframes with gtfparse/pyranges, and sanitizing malformed files with AGAT. Covers the 1-based-inclusive vs 0-based BED coordinate conversion (start-1 only), deriving implicit features (introns/UTRs/TSS), phase-not-frame, the stop-codon-in-or-out-of-CDS convention, and the chr1-vs-1 seqid and gene-ID-version mismatches that silently produce all-zero count matrices and dropped joins. Use when extracting features or sequences from an annotation, converting GTF<->GFF3 or GTF->BED, traversing the gene tree, or diagnosing a coordinate/provenance mismatch upstream of counting or DE.

gptomics Bundle

Tests which individual taxa differ between groups on an amplicon ASV/feature table (phyloseq) using compositionally-aware methods - ALDEx2 (Dirichlet-MC CLR, conservative), ANCOM-BC2/ANCOMBC (sampling-fraction bias correction, structural zeros, passed_ss, default p_adj_method=holm), MaAsLin2/MaAsLin3 (multivariable GLM, random effects, prevalence/abundance split), LinDA (CLR mixed-model regression), ZicoSeq (permutation FDR), LEfSe, and q2-composition ancombc. Covers why the hit list depends more on the DA tool than the biology (Nearing benchmark) so the deliverable is a CONSENSUS of >=2 tools, why a relative change is not absolute without a load anchor, the prevalence-filter knob, BH/FDR plus an effect-size floor, and why DESeq2/edgeR misfire here. Use when finding differentially abundant taxa, handling covariates or longitudinal designs, or choosing a method. Whole-community diversity -> diversity-analysis; shotgun DA -> metagenomics/metagenome-visualization; CoDA theory -> metagenomics/abundance-estimation

gptomics Bundle

Tests for differentially abundant proteins between conditions with limma/DEqMS empirical-Bayes moderation, proDA/msqrob2/MSstats missingness modeling, and Python Welch+BH alternatives. Frames missing values as left-censored MNAR (model, do not impute), makes variance moderation the load-bearing step at n=3-5, and prefers feature/peptide-level testing. Use when identifying proteins with significant abundance changes between experimental groups. Summarization and normalization mechanics are proteomics/quantification; volcano and MA plots are data-visualization/volcano-and-ma-plots; pathway enrichment of the hit list is pathway-analysis/go-enrichment.

gptomics Bundle

Peptide-spectrum matching from MS/MS with target-decoy FDR control, framing identification confidence as a property of a ranked list (q-value/PEP) rather than a raw engine score (XCorr, hyperscore, Andromeda, SpecEValue). Covers sequence-database search engines (Comet, MS-GF+, MSFragger, Sage, MaxQuant, MetaMorpheus), concatenated vs separate target-decoy competition, PEP vs q-value, the multi-level FDR cascade, open/mass-tolerant search, rescoring (Percolator, mokapot, MS2Rescore), and pyOpenMS SimpleSearchEngineAlgorithm + FalseDiscoveryRate. Use when identifying peptides from tandem mass spectra and deciding what FDR threshold to act on. Protein grouping and protein-level FDR are protein-inference; PTM site localization is ptm-analysis; DIA peptide-centric scoring is dia-analysis; intensity quant is quantification.

sboghossian-mini-claude-for-legal Skill

Use when a user on a free or lower tier encounters a capability that requires an upgrade — deep research, higher query limits, eFirm features, Word plugin, or business-plan features. Surfaces a specific, value-led upgrade prompt tied to the exact feature the user just attempted. Governs timing, frequency, and copy so that upgrade prompts feel helpful rather than intrusive. Calibrated against the user's demonstrated value (high-value session = stronger prompt) and emotional state (distress = no prompt).

sboghossian-mini-claude-for-legal Skill

Use when designing the context-caching layer for a legal AI product built on Claude or similar LLMs. Defines how to construct cache keys that maximize prefix-cache hit rates, how to partition context by scope (system prompt, skill, matter, user), how to handle cache invalidation when legal content changes, and how to measure cache efficiency. Engineering skill with significant cost and latency implications for legal AI deployments.

sboghossian-mini-claude-for-legal Skill

Use when building or reviewing the cost accounting layer for a legal AI product — tracking LLM API spend at the granularity of individual requests, skills, matters, users, and tenants. Defines the cost record schema, aggregation dimensions, alerting thresholds, and the connection to the billing model (BYO key vs. platform key). Engineering skill relevant to any Claude-based legal AI deployment.