AI & ML

gptomics Bundle

Measures geometric properties of protein structures with Biopython Bio.PDB - interatomic distances, distance matrices, bond and dihedral angles (phi/psi/chi, Ramachandran), superposition and RMSD, center of mass, radius of gyration, and solvent accessible surface area (SASA). Use when deciding that RMSD depends on BOTH the superposition and the atom selection (a global all-atom RMSD is dominated by flexible loops and hinge motion and is NOT a cross-protein similarity metric); choosing the metric that matches the question (RMSD for same-molecule displacement, TM-score for same-fold, lDDT for superposition-free local model quality - the quantity pLDDT predicts); recognizing Superimposer needs an equal-length ordered atom-to-atom correspondence; and reporting SASA only alongside its probe radius (1.4A water, Shrake-Rupley) with a preference for relative SASA. Keywords RMSD, TM-score, lDDT, SASA, Shrake-Rupley, superposition, Kabsch, dihedral, Ramachandran, radius of gyration.

freedomintelligence Bundle

Estimate causal effects between exposures and outcomes using genetic variants as instrumental variables with TwoSampleMR. Implements IVW, MR-Egger, weighted median, and MR-PRESSO methods for robust causal inference from GWAS summary statistics. Use when testing whether an exposure causally affects an outcome using genetic instruments.

ericgrill Bundle

This skill should be used when designing multi-agent systems that need context isolation, supervisor or swarm coordination, explicit handoffs, parallel execution, or a decision on whether multiple agents are justified.

ericgrill Bundle

Use when you need to SSH into a remote machine to run commands or open an interactive shell using credentials loaded from a .env file (SSH_HOST, SSH_USER, and key or password), avoiding interactive password prompts.

ericgrill Bundle

Design and evaluate context compression strategies for long-running agent sessions. Use when agents exhaust memory, need to summarize conversation history, or when optimizing tokens-per-task rather than tokens-per-request.

ericgrill Bundle

Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.

ericgrill Bundle

This skill should be used when the user asks to "optimize a prompt", "improve prompt performance", "design a prompt template", "write better prompts", "debug prompt issues", "use chain-of-thought", "structured prompting", "few-shot prompting", or wants to apply advanced prompt engineering patterns for production LLM applications.

autohandai-community-skills Bundle

Inspects sandbox health, traces agent behavior, and diagnoses problems. Use when monitoring a running sandbox, debugging agent issues, or checking sandbox logs. Trigger keywords - monitor nemoclaw sandbox, debug nemoclaw agent issues.

autohandai-community-skills Bundle

Translates a HuggingFace model into a prefill-only AutoDeploy custom model using reference custom ops, validates with hierarchical equivalence tests.

sboghossian-mini-claude-for-legal Skill

Use when migrating a PDF document-processing skill originally built for the Anthropic Claude API into the mini-claude-for-legal format. The adapter maps legacy PDF extraction configuration — text layer parsing, OCR fallback, table extraction, form field reading, and annotation handling — into the standard skill model. Critical for legal workflows involving court filings, regulatory submissions, notarised documents, and scanned contracts.

sboghossian-mini-claude-for-legal Skill

Use when drafting an AML/KYC policy for a fintech company or financial institution, covering customer identification, risk assessment, ongoing monitoring, suspicious activity reporting, record-keeping, and regulatory filing obligations. Applies to MENA jurisdictions (UAE, KSA, EG, DIFC, ADGM) and globally, with particular attention to FATF 40 Recommendations and jurisdiction-specific AML regimes.

sboghossian-mini-claude-for-legal Skill

Use to determine whether RAG (Retrieval-Augmented Generation) retrieval should be skipped for a given request. RAG retrieval is the primary source of latency in legal AI chat responses; skipping it for requests that do not need it — generic legal questions, boilerplate drafting, chitchat, admin queries — reduces first-token latency by 600–900ms at p95. Defines precise skip and no-skip conditions with rationale for each.

sboghossian-mini-claude-for-legal Skill

Use when personal data in user-uploaded documents or query content must be redacted before indexing to the vector store or before sending to a third-party LLM provider. Defines the full redaction taxonomy (names, national IDs, IBANs, phones, emails, addresses, health data), placeholder naming conventions, the tenant-scoped reverse-map for rehydration, and the opt-out mechanism for eFirm matters requiring real identifiers. Core to GDPR/PDPL data-minimization obligations and bar-rules confidentiality.

sboghossian-mini-claude-for-legal Skill

Use when running the employment law benchmark that tests drafting, review, calculation, and advisory capabilities across LB/KSA/UAE/DIFC/UK/EU jurisdictions. Contains 30 prompts spanning contracts, termination, EOSG/EOSA calculations, non-compete analysis, and bilingual Arabic requests.

sboghossian-mini-claude-for-legal Skill

Use when migrating a contract-review skill or workflow originally built for Anthropic's Claude API into the mini-claude-for-legal skill format. This import adapter maps the legacy Anthropic contract-review data shape — prompt templates, clause flags, risk buckets, and output schemas — into the standard skill model. Triggers on any import of Anthropic-native contract analysis configurations.

sboghossian-mini-claude-for-legal Skill

Use when importing a DOCX document-processing skill originally authored for the Anthropic Claude API into the mini-claude-for-legal skill format. The adapter handles extraction of text, tables, tracked changes, and metadata from Word documents, mapping the legacy Anthropic processing configuration to the standard skill model with a dry-run preview before committing.

sboghossian-mini-claude-for-legal Skill

Use when migrating a French-law dismissal notification (lettre de licenciement) drafting or review skill into the mini-claude-for-legal format. The adapter maps legacy French employment-law logic — cause réelle et sérieuse requirements, mandatory content checklist, LRAR delivery rules, and CPH (Conseil de Prud'hommes) exposure analysis — into the standard skill model. Primary jurisdiction France; also relevant for Lebanon and French-influenced OHADA labour contexts.

sboghossian-mini-claude-for-legal Skill

Use when migrating a PowerPoint (PPTX) document-processing skill originally built for the Anthropic Claude API into the mini-claude-for-legal format. The adapter maps legacy slide-extraction configuration — text per slide, speaker notes, embedded tables, charts, and metadata — into the standard skill model. Relevant for legal teams processing board presentations, regulatory submissions in slide format, due-diligence summaries, and deal-team pitch decks.

sboghossian-mini-claude-for-legal Skill

Use when a company needs to draft or update a Code of Conduct (or Business Ethics Policy) covering ethical principles, conflicts of interest, gifts and entertainment, anti-bribery and corruption, confidentiality, fair dealing, compliance with laws, and reporting and enforcement mechanisms. Relevant across MENA (UAE, KSA, LB, EG), DIFC/ADGM, and internationally; must address both local anti-corruption law and applicable extraterritorial regimes (UK Bribery Act, US FCPA) for companies with international exposure.

autohandai-community-skills Bundle

Structured framework for verifying numerical parity of HF-to-MCore weight conversions. References existing tools and the add-model-support skill.

sboghossian-mini-claude-for-legal Skill

Use when designing, building, or discussing the memory architecture for a legal-AI agent — including user memory (persistent preferences and context), agent memory (matter-level state, cross-session recall), embedding stores for RAG, recall and forget controls, and the privacy obligations that govern what may be retained. Reach for this skill when the user asks about AI agent memory, persistent context, embedding-based recall, or user data retention design for a legal-AI product.

gptomics Bundle

Calculates molecular fingerprints (ECFP/Morgan, FCFP, MACCS, RDKit, AtomPair, TopologicalTorsion, Avalon, MAP4, MHFP6) and physicochemical descriptors (Lipinski, QED, TPSA, Crippen LogP, 3D shape) with explicit choice tables, bit vs count semantics, and partial-charge model selection. Use when featurizing molecules for similarity, QSAR, virtual screening, or ML, or selecting the correct fingerprint for a chemotype-aware task.

gptomics Bundle

Estimates time-scaled phylogenies, molecular-clock rates, effective reproduction number R_e, and population dynamics from dated pathogen genomes using TreeTime (maximum-likelihood) and BEAST2 (Bayesian; strict/relaxed clocks; coalescent, Bayesian-Skyline, Skygrid, Birth-Death-Skyline, and sampled-ancestor priors; structured coalescent via MASCOT). Covers root-to-tip clock QC via TempEst, date-randomisation tests, recombination masking via Gubbins/ClonalFrameML before clock inference for recombining bacteria, BDSKY origin-vs-rootHeight pitfalls, sampling-bias correction, multi-chain convergence diagnostics, and reconciling phylodynamic R_e with case-based R_t. Use when dating outbreak origins, estimating substitution rates, inferring R_e through time, building time-calibrated Nextstrain Augur trees, choosing between strict and relaxed clocks, fitting Birth-Death-Skyline models, diagnosing temporal-signal failure, running MASCOT for structured-population analyses, or using UShER for pandemic-scale placement.

gptomics Bundle

Infer gene regulatory networks from bulk or general expression data with mutual-information (ARACNe) and tree-ensemble (GENIE3, GRNBoost2) methods, and infer transcription-factor protein activity from regulons with VIPER and msVIPER. Covers the activity-not-edges paradigm, the undirected-association caveat, the DREAM5 wisdom-of-crowds and method-complementarity result, AUPRC-over-AUROC evaluation, and gold-standard incompleteness. Use when inferring a regulatory network from a bulk expression matrix, finding master regulators, or scoring TF activity from a signature. For single-cell motif-pruned regulons see scenic-regulons; for co-expression modules see coexpression-networks.

sboghossian-mini-claude-for-legal Skill

Use when a lawyer submits a deposition transcript, witness statement, sworn declaration, or affidavit for review, and there is any question about whether the testimony may be AI-generated, fabricated, or subject to coordinated coaching. Applies style heuristics, LLM-generation tells, cross-document consistency checks, and implausibility markers to flag specific anomalies — but never labels testimony as fake. Routes high-stakes cases to forensic linguists.

sboghossian-mini-claude-for-legal Skill

Use when drafting a software license agreement for a commercial software product — covering license grant, license type (perpetual/subscription/term), user-definition model, updates and maintenance, IP indemnification, liability cap, and open-source disclosure. Applies to both on-premise and SaaS deployments. Includes attention to MENA-specific issues (data localization, PDPL compliance, SaaS regulation). Triggers on "software license", "saas agreement", "end user license", "EULA", or "software subscription" requests.

sboghossian-mini-claude-for-legal Skill

Use when running the real estate law benchmark across UAE/KSA/Lebanon covering drafting, review, compliance (RERA/Ejari/Tawtheeq/Ejar), disputes, and property tax questions. Contains 30 prompts that test jurisdiction-specific rules including foreign ownership restrictions, RETT, and VAT on commercial rentals.

sboghossian-mini-claude-for-legal Skill

Use when migrating a meeting-briefing skill originally built for the Anthropic Claude API into the mini-claude-for-legal format. The adapter maps legacy meeting-preparation logic — agenda analysis, participant profiling, key-issue summaries, and action-item tracking — into the standard skill model for use in legal and deal-team contexts.

autohandai-community-skills Bundle

Code style guidelines for NeMo-RL (Python and shell). Covers naming, indentation, comments, docstrings, reflection avoidance, and uv usage.

autohandai-community-skills Bundle

Debug AutoDeploy accuracy regressions vs a reference score (PyTorch backend or published baseline). Use when an AutoDeploy model's eval score is significantly below the reference and the root cause is unknown.

sboghossian-mini-claude-for-legal Skill

Use when designing or evaluating pricing strategy for a legal-AI product — covering tiered and usage-based pricing, feature-gating, the free-tier and BYO-key model, enterprise vs SMB pricing dynamics, and the specific considerations of pricing in MENA markets where willingness-to-pay and purchasing structures differ from US/EU SaaS norms. Reach for this skill when the user asks about pricing tiers, monetisation strategy, feature flags, or how to price a legal-AI product.

sboghossian-mini-claude-for-legal Skill

Use when any participant in the system — lawyer, developer, or AI — needs a precise shared definition of a legal-AI term. This reference pack covers the intersection of AI engineering vocabulary (prompts, context windows, RAG, agents, evals) and legal-practice concepts (skills, routers, hallucination, rubrics), enabling unambiguous communication across the two disciplines. Route here when a term appears in a skill description, onboarding document, or internal discussion and its meaning is disputed or unclear.

autohandai-community-skills Bundle

Red team engagement planning is the foundational phase that defines scope, objectives, rules of engagement (ROE), threat model selection, and operational timelines before any offensive testing begins.

autohandai-community-skills Bundle

Extract embedded configuration from Agent Tesla RAT samples including SMTP/FTP/Telegram exfiltration credentials, keylogger settings, and C2 endpoints using .NET decompilation and memory analysis.

autohandai-community-skills Bundle

Bump a pinned dependency (TransformerEngine, Megatron-LM, NRX, etc.), regenerate the lockfile, open a PR, and drive it to green by attaching a watchdog to the "CICD NeMo" workflow and quarantining failing functional tests as flaky until the run is green.

autohandai-community-skills Bundle

Run, monitor, analyze, and debug LLM evaluations via nemo-evaluator-launcher. Covers running evaluations, checking status and live progress, debugging failed runs, exporting artifacts and logs, and analyzing results. ALWAYS triggers on mentions of running evaluations, checking progress, debugging failed evals, analyzing or analysing runs or results, run directories or artifact paths on clusters, Slurm job issues, invocation IDs, or inspecting logs (client logs, server logs, SSH to cluster, tail logs, grep logs). Do NOT use for creating or modifying evaluation configs.