LLM Skills
Table of Contents
If your team writes or maintains QReserve SNEQ scripts, the sneq skill is useful because it helps an AI assistant generate, debug, and review scripts using the actual rules of sneqsneq rather than guessing with Python-like syntax. That matters because sneq is a strict language with its own semantics, return contracts, and failure modes, so a dedicated skill reduces invalid output and makes the assistant more reliable for reservation, rate, approver, invoice, workflow, and form-related scripting tasks.
LLM skills are reusable instruction packages that help an AI assistant perform specific tasks more reliably within a defined domain. A skill can include guidance, workflows, examples, reference material, and supporting files so the assistant has better context for a particular kind of work.
What They Are Used For
LLM skills are used to:
- Improve consistency for repeatable tasks
- Capture domain-specific knowledge and workflows
- Reduce prompt repetition by packaging common instructions
- Help assistants produce more useful first drafts
- Make it easier to version, review, and improve task-specific guidance over time
Common AI Platforms
Many of the most widely used AI platforms can use a skill, even if they call it something different such as custom instructions, project instructions, gems, or agent instructions.
- ChatGPT: Skills can be used as custom instructions or packaged into a custom GPT with instructions, knowledge, and capabilities.
- Claude: Skills can be added as project instructions and paired with project knowledge so Claude uses them across focused chats.
- Google Gemini: Skills can be adapted into a Gem so Gemini follows a reusable set of instructions for a specific task or role.
- GitHub Copilot: Skills can be stored as repository or personal custom instructions so Copilot applies them while coding.
- Microsoft Copilot Studio: Skills can be translated into agent instructions that guide how an agent responds and uses its configured tools and knowledge.
Important Warning
These skills are assistance tools, not guarantees of correctness. They may be incomplete, may miss edge cases, and may not fully reflect the latest behaviour, APIs, product changes, or domain rules. Treat skill output as a starting point that must be reviewed, tested, and validated for the current task and environment.
Use these skills to speed up drafting, analysis, and iteration, but not as a source of truth. Final responsibility for correctness, safety, and production readiness stays with the person using them.
Repository
The LLM skills repository is available here: