Skills for your AI
Analyzes insurance portfolio risk data to produce claims analyses, risk models, scenario and stress tests, compliance briefs, allocation recommendations, and stakeholder reports. Use when the analyst needs portfolio risk insights, model validation, regulatory
Analyzes financial statements, valuations, risk, portfolios, ROI, and investment strategy, and tracks performance. Use when the user provides financial statements or portfolio data, asks for a company valuation, risk assessment, industry research, ROI or cash
Brainstorm product positioning ideas differentiated from competitors. Identifies top competitors and generates positioning statements with rationale. Use when developing product positioning, differentiating from competitors, or crafting brand positioning strat
Gathers, analyzes, and documents post-market surveillance data on medical devices and pharmaceuticals and drafts reports, communications, and study protocols. Use when handling adverse events, complaints, quality monitoring, signal detection, risk assessments,
/cs:post-mortem <decision> — Honest retrospective on an executed decision, scored against original assumptions and dissent. Closes the strategic sprint loop. Use when a decision hits its 90-day review checkpoint or its kill criteria trigger — e.g. scoring last
Turns completed project data into evidence-based lessons-learned reports, performance analyses, dashboards, and closure documentation. Use when a project has finished and the user needs data summaries, objective and deliverable reviews, risk and schedule analy
Turns post-project review data into interview questions, analyses, reports, lessons learned repositories, and action plans. Use when preparing stakeholder interviews, analyzing project documentation, metrics, timelines, risks, or feedback, compiling lessons le
Handles post-sale customer service tasks for sales managers, including follow-up messages, troubleshooting guides, warranty reminders, return guidance, order tracking, surveys, loyalty and referral content, upsell recommendations, knowledge base drafts, and tr
Guides GRPO/RL fine-tuning of language models with TRL for reasoning and structured tasks, covering dataset prep, reward functions, configs, training setup, and log diagnosis. Use when the user wants to prepare data for GRPO, design or test reward functions, g
Guides enterprise RL training for large MoE models with miles, covering MoE setup, speculative RL, INT4 QAT, online MTP, cluster config, and troubleshooting. Use when configuring or debugging miles training runs, checking model/hardware compatibility, or expla
Generates OpenRLHF training commands for PPO, GRPO, and DPO on large models with Ray and vLLM, and troubleshoots common training failures. Use when configuring or launching RLHF jobs, choosing between PPO/GRPO/DPO, or fixing OOM, DeepSpeed, instability, or slo
Guides SimPO LLM alignment training by drafting YAML configs, launch commands, troubleshooting fixes, algorithm comparisons, and hardware guidance. Use when a user wants to set up or debug a SimPO run, choose between SimPO/DPO/PPO/GRPO, or size GPUs for prefer
Guides LLM post-training with RL in the slime framework, covering GRPO setup, async training, multi-turn agentic training, config validation, data format, model scripts, and monitoring. Use when a user asks how to set up or run slime training, validate batch-s
Guides PyTorch-native agentic RL experiments with Meta's torchforge, covering GRPO setup, custom loss functions, distributed training configs, and installation. Use when the user asks to set up GRPO training, implement a custom RL loss, configure multi-GPU or
Fine-tunes and aligns language models with TRL methods (SFT, DPO, PPO, GRPO, reward model training) and evaluates the results. Use when the user wants to train on instruction or preference data, run RLHF pipelines, or test an aligned model on prompts.
Guides reinforcement learning post-training of LLMs with the verl library, covering GRPO, PPO, Megatron backend, datasets, installation, algorithm selection, monitoring and troubleshooting. Use when configuring or debugging verl RL training runs, choosing an R
Optimizes PostgreSQL performance, designs high-availability replication, and builds backup and recovery strategies. Use when queries are slow, replication or failover must be designed, backups or RPO/RTO need improvement, autovacuum and bloat need tuning, or a
Manages PostgreSQL databases by connecting, querying, modifying schema, loading CSV data, backing up and restoring, monitoring performance, and visualizing schemas. Use when the user asks to connect to a PostgreSQL server, run or optimize SQL, alter tables, im
/em:postmortem — Honest analysis of what went wrong. Use after a failed launch, missed quarter, or bad hire to run a blameless 5-Whys retrospective with a change register — e.g. dissecting why the Q3 release slipped six weeks.
Guides Power BI data model design using star schema, relationship, storage mode, performance, and security best practices from Microsoft documentation. Use when structuring fact and dimension tables, fixing relationship filtering, choosing Import/DirectQuery/C
Provides expert DAX guidance for Power BI formulas, performance, error handling, time intelligence, calculation groups, and DirectQuery optimization using Microsoft best practices. Use when writing or reviewing DAX measures, fixing slow or wrong results, or de
Optimizes Power BI model, report, DAX, DirectQuery, composite, and capacity performance using Microsoft best practices. Use when a user reports slow reports, slow DAX measures, DirectQuery or composite model issues, or high Premium capacity utilization.
Guides Power BI report design and visualization using Microsoft best practices for chart selection, layout, interactivity, performance, accessibility, and design patterns. Use when planning or improving a Power BI report, dashboard, or visual.
Analyzes power grid data for load flow, faults, stability, renewables, modernization, transmission, distribution, resilience, expansion, microgrids, cybersecurity, and asset management, returning structured findings and recommendations. Use when an energy engi
