AI-SKILL

Review code changes and report issues by severity with actionable fixes.

Sharpen a fuzzy intention into one measurable objective string that drives the rest of the work.

Convert a Prompt Flow PRS pipeline submission to run a Microsoft Agent Framework workflow.

Build a Model Context Protocol (MCP) server that lets an LLM call into external tools and resources.

Summarize PDF documents into concise bullet-point digests.

Bump a dependency version across a pnpm workspace and update lockfile.

Convert Prompt Flow flow.dag.yaml definitions into runnable Microsoft Agent Framework workflow code.

Extract invoice numbers, dates, amounts and other fields from scanned PDFs using Claude vision.

Audit a UI or component for accessibility issues and fixes.

Review an API design against consistency, versioning, auth, and error-handling best practices.

Design a versioning and deprecation strategy for an API.

Create a backup and recovery checklist for a datastore or service.

The user wants to run an ML model in JavaScript

The user wants a **real CLI** they can run by name

You're reviewing or designing something that touches

Review code for performance bottlenecks and suggest measured optimizations.

Canonical code-review workflow for any PR, branch, diff, or local change. Findings ordered by severity, defer to specialist skills when needed.

Generate commit messages following Conventional Commits spec, including subject/body/breaking-change footer.

Plan a safe data migration between schemas, formats, or systems.

Review a query or schema for missing, redundant, or misused indexes.

Targeted fix to attempt next round

Suggested follow-up actions

Confirmation that traces are arriving in Confident AI's

Plan and execute a low-risk dependency update for a project.

Respond to a reported CVE in a project dependency.

A three-stage structured document co-authoring workflow - Context Gathering / Refinement & Structure / Reader Testing. Triggered when user mentions writing docs, drafting proposals, or creating specs. Avoid the write-and-forget trap.

Design a documentation outline for a project or feature.

The user wants to train, fine-tune, or adapt an

Draft a language-agnostic error-handling strategy for a module or service.

Bootstrap a minimal evaluation suite for an LLM-powered feature.

Design a feature-flag rollout strategy for a new feature.

You're about to build a **non-trivial frontend feature** —

The user is building something where the quality

Draft a blameless postmortem from an incident timeline.

You're drafting a **piece of internal communication** —

You maintain an internal **per-token LLM pricing data file** —

Define a structured logging convention for a service or library.

Evaluate whether a proposed service boundary is coherent and low-coupling.

Define metrics, dashboards, and alerts for a service.

Validates and repairs identifier naming conventions in codebases.