Total 56,626 skills, AI & Machine Learning has 9429 skills
Showing 12 of 9429 skills
OpenRouter unified AI API - Access 200+ LLMs through single interface with intelligent routing, streaming, cost optimization, and model fallbacks
Production Brief methodology for complex Veo 3 video scenes. Use when creating scenes with dialogue, character continuity, structured settings, or multi-beat sequences. Provides 11-block framework (Format & Tone, Main Subjects, Wardrobe & Props, Location & Framing, Lighting & Palette, Continuity Rules, Actions & Camera Beats, Montage Plan, Dialogue, Sound & Foley, Finish) for professional, replicable results.
Strategy for creating efficient short-form video prompts. Use when creating filler shots, atmospheric scenes, or quick video clips that don't require full Production Brief methodology. Covers when to go short vs long, format+style upfront rule, and two approaches (Descriptive vs Directive) for compact yet coherent results.
Cancel any active OMC mode (autopilot, ralph, ultrawork, ecomode, ultraqa, swarm, ultrapilot, pipeline, team)
Create a SageMaker endpoint (real-time or async) with autoscaling, CloudWatch alarms, and tagging enabled by default. Use this skill whenever about to create a SageMaker endpoint, write deployment code that calls `create_endpoint`, or finalize a deployment after the image URI and IAM role are known. Provides deploy.py for real-time endpoints and deploy_async.py for async endpoints (with genuine scale-to-zero support). This is the last step in the SageMaker deployment workflow. Never generate a bare `create_endpoint` call without these defaults — endpoints without autoscaling or alarms are demos, not deployments.
AI SDLC host adapter and capability negotiation workflow. Use when an AI assistant needs to validate a host adapter manifest, map portable workflow operations to host-native operations, negotiate capabilities and limits, select deterministic semantic-preserving fallbacks, or explain why a host cannot run a plan. Supports `--quick-flow` and `--full-flow`.
When facing legal disputes, case consultations, or preparing litigation strategies, this skill is used to clarify case facts, organize legal relationships, formulate defense arguments, and compile evidence lists. After the report is completed, it proactively provides a follow-up analysis menu (precedent search, element subsumption, relationship diagram, compliance review, text revision) for users to choose and refer to.
Guide the Agent to consolidate case facts, legal provisions, court judgments, issues, parties, evidence, contract obligation relationships (Three-layer Model of Contract → Clause → Obligation, including Risk Rating Color Coding), and constitutive element subsumption results (Law → Element → Fact Subsumption Color Coding), generate superset legal relationship graph data compatible with law-powers' index.html/data.js, and write it into data.js.
Evaluates accuracy of quantized or unquantized LLMs using NeMo Evaluator Launcher (NEL). Triggers on "evaluate model", "benchmark accuracy", "run MMLU", "evaluate quantized model", "accuracy drop", "run nel". Handles deployment, config generation, and evaluation execution. Not for quantizing models (use ptq) or deploying/serving models (use deployment).
Representative MoE training playbooks by hardware platform and model family. Summarizes rounded throughput bands, parallelism patterns, and common tuning stacks.
Manage durable working-session memory for coding agents. Use when a user asks to preserve or recover agent context across disconnects, VS Code restarts, long-running work, handoffs, or any session where important state should be written periodically under the repo's session directory.
External verl end-to-end validation workflow for Megatron-Bridge model/provider changes. Covers running a small verl Megatron backend job from a Bridge checkout, choosing LoRA/DDP plus optional save/resume and parallelism variants, setting PYTHONPATH so verl imports the local Bridge tree, and reporting pass/fail evidence.