agentic▌

Iterative evaluation and refinement patterns for improving AI agent outputs through self-critique loops. \n \n Provides three core patterns: basic reflection (self-critique loops), evaluator-optimizer (separated generation and evaluation), and code-specific test-driven refinement \n Supports multiple evaluation strategies including outcome-based assessment, LLM-as-judge comparison, and rubric-based scoring with weighted dimensions \n Includes practical Python implementations with structured JSON

Framework for effective AI collaboration defining task decomposition, context management, abstraction selection, and automation philosophy. \n \n Divide complex tasks into small, independently verifiable steps; AI performs significantly better with clear, bounded instructions than large ambiguous requests \n Keep context fresh and single-purpose; use separate conversations for different topics and apply HANDOFF.md summaries when conversations grow long to prevent context drift \n Adjust abstract

Core principles for effective AI agent collaboration: divide tasks, manage context, choose abstraction levels, automate patterns, and validate results. \n \n Decompose large tasks into small, explicit steps; AI performs 39% better on focused instructions than vague, complex requests \n Keep context fresh and single-purpose using separate conversations or HANDOFF.md documentation to avoid context drift and performance degradation \n Switch between high-level \"vibe coding\" for prototyping and de

AWS Bedrock AgentCore provides a complete platform for deploying and scaling AI agents with seven core services. This skill guides you through service selection, deployment patterns, and integration workflows using AWS CLI.

Transform a full-stack app into one where agents are primary actors — reasoning, coordinating, remembering, and communicating alongside humans.

Load with: base.md + llm-patterns.md + [language].md

Notice: This repository is experimental and subject to change without notice. By using the features and skills in this repository, you agree to Alchemy's Terms of Service and Privacy Policy.

Standard multi-agent pipeline for implementation tasks.

Static security analysis for GitHub Actions workflows invoking AI coding agents. \n \n Detects nine attack vectors where attacker-controlled input reaches AI agents in CI/CD pipelines, including env var intermediaries, direct expression injection, CLI data fetches, dangerous sandbox configs, and wildcard user allowlists \n Scans .github/workflows/ for Claude Code Action, Gemini CLI, OpenAI Codex, and GitHub AI Inference steps; resolves one level of composite actions and reusable workflows \n Cap

AI-driven engineering workflows with eval-first execution, task decomposition, and cost-aware model routing. \n \n Defines an eval-first loop: establish baseline evals before implementation, then re-run post-execution to measure deltas and catch regressions \n Decomposes work into 15-minute units with single dominant risks, independent verifiability, and clear done conditions \n Routes tasks by complexity: Haiku for classification and boilerplate, Sonnet for implementation, Opus for architecture