- Home
- Skills
- Maxritter
- Pilot Shell
- Standards Skill Creator
standards-skill-creator_skill
- TypeScript
413
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill maxritter/pilot-shell --skill standards-skill-creator- SKILL.md5.6 KB
Overview
This skill is a practical guide for creating and maintaining Claude Code skills that add domain knowledge, workflows, and tool integrations. It focuses on concise context, appropriate degrees of freedom, and reusable bundled resources so skills stay efficient and reliable. Use it when planning, building, or updating a skill that will be loaded into Claude's context.
How this skill works
The guide inspects the intended task surface and recommends a minimal set of metadata, concise instructions, and optional bundled resources (scripts, references, assets). It defines when each resource should be loaded into Claude's context and prescribes a three-stage disclosure model: metadata, main instructions, and on-demand resources. It also provides naming conventions, testing steps, and a checklist to ensure skill quality.
When to use it
- Creating a new Claude skill to automate a multi-step workflow or integrate an API
- Updating an existing skill to reduce token cost or improve reliability
- Designing skills that require deterministic code or domain-specific references
- Packaging reusable scripts, templates, or large documentation for on-demand loading
- Preparing skills for production with automated testing and clear triggers
Best practices
- Keep context terse: include only facts Claude likely doesn’t already know
- Choose freedom level to match task fragility (high/medium/low)
- Provide concise examples instead of long explanations
- Organize reusable items as scripts, reference docs, or output assets
- Expose only metadata in initial context; load heavy docs on demand
- Test skills with real use cases and iterate based on failures
Example use cases
- A PDF processing skill with short code snippets and a reference for advanced forms
- A BigQuery skill that loads only the relevant schema file for the current query
- A deployment helper that bundles deterministic scripts and templates for CI
- A finance-reference skill that provides compact triggers and loads large docs when needed
- An API-integrator skill that exposes a few safe parameters and precise scripts for fragile calls
FAQ
Include a concise description of what the skill does and all trigger contexts; keep it short so it fits comfortably in the initial context.
When should I add bundled reference files?
Add references when documentation is large or frequently consulted; load only the specific reference file relevant to the current task.