How It Works
The otc-awesome-llm collection gives your AI coding assistant superpowers. It's built from 4 types of building blocks and 3 orchestration plugins that work together across VS Code, Claude Code, and Codex CLI.
The 4 Building Blocks
Every asset in the collection is one of these four types.
Prompts
One-shot instructions you paste into any AI chat. Each prompt targets a specific task like generating Terraform modules or writing unit tests.
"Generate a Terraform module for an Azure Key Vault with private endpoint."
BrowseInstructions
Persistent coding standards that stay active across your entire session. They shape how the AI writes code — naming conventions, error handling patterns, and more.
"Always use structured logging with correlation IDs."
BrowseAgents
Multi-step specialists that own an entire workflow. An agent can scaffold a project, run tests, fix failures, and commit — all autonomously.
"Scaffold a new microservice with CI pipeline, Dockerfile, and Helm chart."
BrowseSkills
Deep domain knowledge modules that teach the AI about a specific technology or practice. Skills provide context, not commands.
"Understand FHIR R4 resources, HIPAA safeguards, and healthcare data patterns."
BrowseHow They Compose
These building blocks aren't isolated — they layer together to give your AI assistant deep context and precise capabilities.
Instructions form the base layer; skills add domain context on top
Composition is user-driven
You choose which assets to load for each session. Working on Terraform? Load the Azure skill and your team's IaC coding standards. Debugging a pipeline? Swap in the CI/CD skill and a troubleshooting prompt.
Mix and match freely
Assets are independent and composable. Any prompt works with any instruction. Any agent can leverage any skill. There are no required bundles — use one asset or twenty, whatever fits your workflow.
Example: Debugging a K8s CrashLoopBackOff
Here's how the four building blocks combine in a real debugging scenario.
- 1Skill
Load the Kubernetes skill so the AI understands your cluster topology and Helm patterns.
- 2Instruction
Activate your team's coding standard instruction — structured logging, error handling, naming conventions.
- 3Prompt
Paste the "Debug K8s CrashLoopBackOff" prompt to kick off targeted troubleshooting.
- 4Agent
The K8s debugging agent takes over — inspects pods, reads logs, patches the deployment, and verifies the fix.
The 3 Plugins
Plugins are higher-level orchestrators built on top of the building blocks. Each one answers a different question.
Professor Hudak
“What exists? What should I use?”
Searches the entire otc-awesome-llm repository and recommends the right prompts, instructions, agents, and skills for your current task.
AI-DLC
“Help me build something new.”
Guides you through a structured development lifecycle — from requirements gathering through design, implementation, and testing.
Dr. Zero
“Improve this repo autonomously.”
Autonomous repository improvement engine. Analyzes CI failures, code quality issues, and infrastructure gaps, then fixes them with a swarm of specialist agents.
Finding the Right Asset
Every asset is tagged with technologies, domains, and use cases. Use tags to narrow down exactly what you need.
Ready to get started?
Set up your IDE in minutes and start using 132+ curated AI assets.