accelerate-data

Use when reviewing a vd-studio pull request as the automated AI PR reviewer (OpenHands pr-review plugin) — judging test coverage for a diff, structuring the review report, or deciding merge readiness.

Use when test quality decisions are needed for new, changed, brittle, missing, or review-requested unit or component tests, including tests-first work, assertion choice, mock boundaries, lifecycle scenarios, or independent test-session handoff.

Use when filing, logging, raising, reporting, or submitting a product issue, feature request, bug, or feedback for the Studio or Utilities product areas in Linear.

Grilling session that challenges your plan against the existing domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as decisions crystallise. Use when user wants to stress-test a plan against their project's language and documented decisions.

Compact the current conversation into a handoff document for another agent to pick up.

Use when asked to implement, fix, build, or work on an existing Linear issue in this repository, and coding should stop before PR creation

Use when the user wants to add a roadmap item, submit a roadmap request, propose a product roadmap candidate, or log RO-team roadmap intake in Linear. Not for generic Linear issue creation, bug reports, or implementation tickets.

Use when the user says "close issue [ID]", "close [ID]", "finish closeout", or when a Linear issue already has a review-ready PR and this repository needs the merge, close, and cleanup phase

Use automatically before any completion, successful, passing, PR-ready, merged, or stage-complete statement to verify fresh evidence and choose accurate claim wording.

Use internally when /refactor-mart dispatches one approved Type: int or Type: mart candidate

Use when a monolith warehouse or lakehouse has a whole-warehouse `warehouse-ddl/` snapshot and needs tables or views grouped into business data domains with data owners for data mesh decentralization, setup-source selection, migration planning, or scope and migrate-mart inputs.

Use when generating or revising one dbt model for a single profiled table or view after refactor and approved test-spec work are complete.

Use when the generate-model workflow needs a read-only review of generated dbt model artifacts before accepting the result or sending it back for fixes.

Use internally when /refactor-mart dispatches one approved Type: stg candidate

Use when a migrated table, view, or materialized view needs a new or merge-safe `test-specs/<schema.object>.json`, especially after SQL changes or reviewer feedback.

Use when a selected mart unit must be analyzed into staging and higher-layer refactor candidates before any code changes

Practitioner-level dbt patterns for Microsoft Fabric. Covers materialization decisions, incremental model gotchas, snapshot strategies, selector/tag patterns, ref() chain rules, and Fabric-specific T-SQL quirks. Use when writing, reviewing, or debugging dbt models on Fabric. Also use when setting up dbt_project.yml configuration or CI/CD pipelines for dbt-fabric projects.

Define semantic models and MetricFlow metrics in dbt on Microsoft Fabric. Use when creating semantic model YAML, defining entities, dimensions, and measures, or configuring metric types. Also use when choosing between denormalized marts and semantic layer queries.

Implement SCD Type 2 snapshots in dbt on Microsoft Fabric. Use when building snapshot models, choosing between timestamp and check strategies, or handling Fabric-specific datetime2 quirks. Also use when tracking slowly changing dimensions or debugging snapshot drift.

Configure Elementary anomaly detection for dbt on Microsoft Fabric. Use when placing data quality tests by medallion layer, setting up volume or column anomalies, or configuring timestamp_column. Also use when debugging silent test failures or choosing severity levels.

Build dlt REST API pipelines to ADLS Gen2 and OneLake. Use when configuring RESTAPIConfig, handling pagination, or setting schema contracts for REST sources. Also use when landing REST API data into Fabric lakehouses or debugging schema evolution.

Use when modeling ASC 606 revenue recognition in dbt on Microsoft Fabric, including performance obligations, transaction price allocation, deferred revenue, contract modifications, or silver/gold revenue reporting layers.

Extract Salesforce data via dlt into dbt on Microsoft Fabric. Use when building Salesforce pipelines, handling CPQ overrides, soft deletes, or managed package fields. Also use when choosing CDC timestamps or debugging missing Salesforce records.

Core Fabric CLI (fab) operations — workspace navigation, item management, authentication, resource types, path format, output formatting. Use when listing workspaces/items, creating/deleting resources, checking existence, describing items, or configuring the CLI.

Accelerate Data visual brand guidance for vibeData UI. Use for brand-compliant premium surfaces across product, marketing, docs, demos, and internal tools when the interface represents Accelerate Data.

Design system component workflow. Spec, document, implement, review, spec-review, and audit DS components with Figma as primary input. 6 actions: spec, doc, dev, review, spec-review, audit. Auto-activates on: "component", "spec component", "doc component", "dev component", "review component", "spec-review component", "audit component", "design system", "ds component", "component spec", "component doc", "component dev", "component review", "component spec-review", "component audit".

Compose UI screens from existing components. Proposes ASCII wireframe layouts with real component mapping and coverage %, then specs the chosen layout. 6 actions: init, spec, spec-review, craft, ship, review. Core guarantee: zero component reinvented. Auto-activates on: "design-screen", "design-screen spec", "design-screen init", "design-screen craft", "design-screen ship", "design-screen build", "design-screen review", "design-screen spec-review", "compose screen", "screen layout", "compose ui", "design screen".

Design system component workflow. Spec, document, implement, review, spec-review, and audit DS components with Figma as primary input. 6 actions: spec, doc, dev, review, spec-review, audit. Auto-activates on: "component", "spec component", "doc component", "dev component", "review component", "spec-review component", "audit component", "design system", "ds component", "component spec", "component doc", "component dev", "component review", "component spec-review", "component audit".

Compose UI screens from existing components. Proposes ASCII wireframe layouts with real component mapping and coverage %, then specs the chosen layout. 6 actions: init, spec, spec-review, craft, ship, review. Core guarantee: zero component reinvented. Auto-activates on: "design-screen", "design-screen spec", "design-screen init", "design-screen craft", "design-screen ship", "design-screen build", "design-screen review", "design-screen spec-review", "compose screen", "screen layout", "compose ui", "design screen".

Use when authoring or updating developer-facing design specs under docs/design for one or more canonical functional specs (many-to-many via tags), especially architecture decisions, technical tradeoffs, states, data flows, schemas, source-file references, and gaps between flow behavior and current design coverage

Use when creating or updating a functional spec for a user flow listed in column B of the User-Flows-Details Sheet

Use when creating or updating end-user help pages, user documentation, screen docs, feature help pages, or docs under `docs/user-guide/`. Trigger on "author user guide", "create user guide", "new user guide", "update user guide", "document <screen>", "help page for <feature>", or "/authoring-user-guide".

Use when the user wants to write or improve a prompt for Claude, ChatGPT, Gemini, Cursor, Claude Code, GitHub Copilot, Windsurf, Bolt, v0, Midjourney, DALL-E, Stable Diffusion, or another AI-powered tool.

Use when writing or tightening prose for functional specs, design documents, user guides, or implementation plans, especially when draft text is wordy, vague, passive, duplicative, or hard to scan. Do not use for AI prompt-writing, code comments, commit messages, general chat, or unrelated prose.

Grade all capstone submissions and produce a ranked comparison table. Use when someone asks to evaluate all submissions, rank all candidates, compare all students, or says: grade all, evaluate all, rank submissions, compare all. Accepts a list of name:url pairs or uses local ext/ paths.

Evaluate a single capstone submission against the rubric. Use when someone provides a student name, git URL, or submission path and wants it graded, scored, or assessed. Trigger phrases: grade arpit, evaluate duy, score ext/khang, assess this submission, how did [name] do. Supports --deep-dive flag for detailed technical analysis.

List all available capstone evaluation rubrics. Use when someone asks what rubrics exist, what they can evaluate against, or says: list rubrics, what rubrics are available, what roles can I evaluate.

List all available screening rubrics. Use when someone asks what rubrics exist, what they can screen for, which rubrics are available, or says: list rubrics, what can I screen, available rubrics, what rubrics do we have.

Screen a resume against a job rubric. Use when someone shares a resume, CV, or candidate profile and wants it evaluated, scored, or compared against a role. Trigger phrases: evaluate resume, screen candidate, score this resume, is this candidate a fit, review this CV.

Creates Amazon-style one-pager documents for product proposals, feature pitches, initiatives, or business cases. Use when the user wants to write a one-pager, create a proposal document, draft a concise executive proposal, or mentions "one-pager", "1-pager", "one pager", or "Amazon-style proposal". Guides the user through structured information gathering and produces a polished narrative document.

Orchestrates a structured multi-agent adversarial debate on any strategic question. Four agents (Maximalist, Purist, Hybrid, Economist) write position papers, debate through rebuttals, and produce a synthesized recommendation. Activates when the user wants to analyze a strategic decision from multiple angles, run a red team analysis, or get a rigorous recommendation on any topic where reasonable people disagree. Triggers on debate, adversarial analysis, red team, argue both sides, tradeoffs, position paper, devil's advocate, or help me think through / decide on a topic with genuine tension.

Use when proposing new features/products, documenting product requirements, creating concise specs for stakeholder alignment, pitching initiatives, scoping projects before detailed design, capturing user stories and success metrics, or when user mentions one-pager, PRD, product spec, feature proposal, product requirements, or brief.

Packages session research into a review-ready folder for co-founders and colleagues. Creates deliverable files (if only conversation exists), a README that IS the proposal, a research-path doc, and a chat history — whether the session produced files or only conversation. Use when the user says "package for review", "share this with the team", "make this shareable", "wrap this up for [name]", "review package", "prep for handoff", "summarize this session for others", "summarize for co-founders", "summarize for cofounders", "summarize for others", "summarize this for [name]", "create a review package", or any variation of wanting to turn session output into something others can read.

Use when the user asks to "write a DAX measure", "create DAX calculations", "add time intelligence", "calculate YTD", "year over year", "running total", "semi-additive measure", "parent-child hierarchy", "ABC classification", "new and returning customers", "dynamic segmentation", "currency conversion", "ranking", "cumulative total", "budget allocation", or any DAX formula authoring for a Microsoft Fabric or Power BI semantic model. This skill provides DAX patterns specifically formatted as TMDL for Fabric semantic models.

Use when the user asks to "design a Fabric semantic model", "create a TMDL model", "build a Fabric star schema", "convert dbt to TMDL", "scaffold a Fabric tabular model", "set up a Fabric dataset", "create a Power BI semantic model", or needs guidance on Microsoft Fabric TMDL file structure, naming conventions, DirectLake configuration, or Fabric semantic model architecture best practices. This skill is specifically for Microsoft Fabric and Power BI semantic models, not general data modeling.

Use when the user asks to "validate my semantic model", "check my TMDL", "review my Fabric model", "audit my Power BI model", "run best practice checks", "lint my TMDL", or wants to verify correctness, naming conventions, relationship integrity, measure quality, or performance of a Microsoft Fabric semantic model. This skill validates TMDL files against Fabric best practices.

Use when an agent needs to find which Vibedata documentation source(s) cover a topic, including questions like "where is X documented in Vibedata", "load context for X", "find docs about Y in vd-intelligence/strategy/architecture/flows", or any question that requires routing across the Vibedata source graph before reading content. Auto-activates on Vibedata-specific routing questions; does not activate for generic web/library docs.

Best practices for Remotion - Video creation in React

Use when writing a new reusable skill from clarified requirements, confirmed decisions, examples, constraints, and expected outputs, including SKILL.md, focused references, and fresh-context verification.

Use when deciding what to research and what clarification questions to ask so user intent, trigger conditions, examples, edge cases, dependencies, and guardrails can inform creation or refinement of a skill.

Use when building or debugging Tauri apps with src-tauri, tauri.conf.json, capabilities, permissions, invoke, events, @tauri-apps/api, Rust commands, plugins, sidecars, WebView2, signing, updater, mobile, build, bundle, or desktop distribution.

Combine multiple mental models for richer analysis. Use for complex problems requiring multiple lenses, high-stakes decisions, or when single models leave blind spots.