accelint-skill-manager_skill

Overview

This skill helps create, refactor, audit, and package skills that extend AI agents with specialized knowledge, workflows, or tool integrations. It is designed for authors who need a repeatable, agent-friendly format and clear triggering descriptions. Use it when adding or maintaining skills that agents will load or compose into workflows.

How this skill works

The skill inspects skill metadata, description fields, triggers, and supporting scripts to ensure they follow agent-friendly conventions and loading heuristics. It validates that descriptions are trigger-focused (start with "Use when"), technology specificity is explicit, and cross-skill references use the required skill-name format. It can scaffold shell-based packaging scripts and surface concrete fixes for description, keyword coverage, and cross-referencing issues.

When to use it

• When creating a new skill for an AI agent ("create a skill", "make a new skill", "build a skill")
• When packaging functionality as a reusable skill or preparing a skill for distribution
• When refactoring or updating an existing skill to improve discoverability or correctness
• When auditing skills to ensure they follow agent-loading and trigger conventions
• When converting ad-hoc scripts or workflows into agent-loadable skills

Best practices

• Always start the description with "Use when" and describe triggering conditions, not the workflow
• Keep triggers concrete and problem-focused (symptoms, error messages, environments), technology-specific only when appropriate
• Cover relevant keywords and error messages agents might search for to improve discoverability
• Reference other skills by exact skill name with explicit requirement markers (e.g., REQUIRED SUB-SKILL: skill-name)
• Provide shell packaging scripts or simple build steps when primary language is Shell to ease installation

Example use cases

• Authoring a new skill that adds a CI test-generation helper and packaging it for agent consumption
• Refactoring a skill whose description summarized workflow so agents incorrectly skipped detailed checks
• Auditing a set of skills to ensure keyword coverage for common error messages like "timeout" or "ENOTEMPTY"
• Converting a collection of shell helpers into a single agent-loadable skill with explicit triggers
• Preparing a skill that integrates with third-party tools and ensuring cross-skill references are correctly marked