kimtth/azure-ml-finetuning-eval-skills
Skill Claude CodeCodex
Train or fine-tune LLMs on Azure ML managed compute with TRL trainers. Uses direct trainer loops (SFT, DPO, RL) without relying on serverless APIs or Hugging Face infrastructure.
kimtth/azure-ml-finetuning-eval-skills
Skill Claude CodeCodex
Evaluate generative AI applications and models locally or in the cloud using Azure AI Evaluation SDK. Measure quality, safety, and performance with built-in and custom evaluators.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: review an academic paper with a structured review form and multiple reviewer personas.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: persist and recall context across coding sessions so the agent stops re-learning the same stack, bugs, and decisions.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: design a clear, consistent, and evolvable API contract before implementing it.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: record a significant technical decision, its context, and consequences as an ADR.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: implement or review authentication flows, sessions, tokens, and password handling securely.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: design or review access control so users can only do what they are permitted to do.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: write or review robust shell scripts that fail fast and handle paths and input safely.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: explore options or decompose an ambiguous problem before committing to one approach.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: add caching with the right layer, key design, invalidation, and consistency trade-offs.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: debug, inspect, automate, screenshot, profile, or analyze a web page with Chrome DevTools-style browser tools.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: design or improve a CI/CD pipeline for fast, reliable, automated build-test-deploy.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: structure an app so business logic is independent of frameworks, UI, and databases.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Clean up dead code, obsolete paths, and contradictions in the codebase to keep it small, correct, and aligned with reality.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: write or review concurrent code that avoids races, deadlocks, and shared-state bugs.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: write or review modern C++ with RAII, smart pointers, and value semantics.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: give structured, specific design critique with actionable fixes, not vague feedback.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: write or review modern, idiomatic C# with async, nullable reference types, and clean structure.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: write and apply schema migrations safely without downtime or data loss.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: decouple components by injecting their dependencies instead of constructing them inline.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: apply a design pattern to a real problem without over-engineering the solution.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: multiple independent coding, debugging, or research tasks can be investigated in parallel.
kimtth/agent-skill-100-lines-or-less
Skill Claude CodeCodex
Use when: define multi-service local and test environments with Docker Compose reliably.