Skill manager

Skill aida-core/aida-core-plugin/skills/skill-manager

Create, validate, version, and list Claude Code skill definition files following the Agent Skills open standard (two-phase API). Use for CRUD on skill files, routed through /aida extension management. Do NOT use to execute a skill's behavior — this only authors and maintains skill definitions.From its SKILL.md

Install

npx -y skills add aida-core/aida-core-plugin --skill skill-manager

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.8 KB, as published. Nobody here has run it

Skill Manager

Focused management interface for Claude Code skill definitions. Skills are process definitions with execution capabilities that follow the Agent Skills open standard.

Activation

This skill activates when:

User invokes /aida skill create [description]
User invokes /aida skill validate [name|--all]
User invokes /aida skill version [name] [major|minor|patch]
User invokes /aida skill list
Skill management operations are needed

Operations

Create

Create a new skill definition using the three-phase orchestration pattern.

Phase 1: Gather Context (Python)

Run the script to infer metadata and get questions:

python {base_directory}/scripts/manage.py --get-questions \
  --context='{"operation": "create", "description": "user description", "location": "user"}'

Returns:

{
  "inferred": {
    "name": "inferred-name",
    "version": "0.1.0",
    "tags": ["custom"],
    "base_path": "~/.claude/skills/inferred-name"
  },
  "questions": [],
  "project_context": {
    "languages": ["python"],
    "frameworks": ["fastapi"]
  }
}

Phase 2: Generate Content (Agent)

Spawn the claude-code-expert agent with complete context:

Operation: CREATE
Type: skill

Requirements:
  - Name: {inferred.name}
  - Description: {user_description}
  - Location: {inferred.base_path}
  - Project Context: {project_context}

Output Format:
Return a JSON object with this exact structure:

{
  "validation": {
    "passed": true|false,
    "issues": [
      {"severity": "error|warning", "message": "...", "suggestion": "..."}
    ]
  },
  "files": [
    {"path": "skills/{name}/SKILL.md", "content": "..."},
    {"path": "skills/{name}/scripts/{script}.py", "content": "..."},
    {"path": "skills/{name}/references/{doc}.md", "content": "..."},
    {"path": "skills/{name}/templates/{tmpl}.jinja2", "content": "..."}
  ],
  "summary": {
    "created": ["list of relative paths"],
    "next_steps": ["actionable items for user"]
  }
}

Phase 3: Write Files (Python)

Pass agent output to Python for file creation:

python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "create", "agent_output": <agent_json>, "base_path": "~/.claude"}'

The script validates file structure, creates directories, writes files, and returns success or failure.

Validate

Validate skill frontmatter against the schema.

# Validate specific skill
python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "validate", "name": "my-skill"}'

# Validate all skills
python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "validate", "all": true, "location": "all"}'

Version

Bump a skill's semantic version.

python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "version", "name": "my-skill", "bump": "patch"}'

List

List all discovered skills.

python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "list", "location": "all"}'

Skill Directory Structure

Skills follow the Agent Skills open standard directory layout:

skills/
  my-skill/
    SKILL.md        # Skill definition (frontmatter + instructions)
    references/     # Workflow documentation
      workflow.md
    scripts/        # Executable scripts
      manage.py
    templates/      # Optional Jinja2 templates
      template.jinja2

Path Resolution

Base Directory: Provided when skill loads via <command-message> tags.

Script Execution:

{base_directory}/scripts/manage.py

Reference Files:

{base_directory}/references/create-workflow.md
{base_directory}/references/validate-workflow.md
{base_directory}/references/schemas.md

Location Options

Location	Path	Use Case
`user`	`~/.claude/`	Personal skills
`project`	`./.claude/`	Project-specific skills
`plugin`	Custom path	Plugin development

Example Workflow

Creating a Skill (Full Flow)

User: /aida skill create "auto-format code on save"

1. Parse: operation=create, description="auto-format code on save"

2. Phase 1 (Python):
   python manage.py --get-questions --context='{...}'
   Returns:
   - inferred: name="auto-format", version="0.1.0"
   - questions: [location question]
   - project_context: {languages: ["python"]}

3. Ask user questions (if any):
   "Where should we create this skill?"

4. Phase 2 (Agent):
   Spawn claude-code-expert with context + output contract
   Agent returns JSON with files array

5. Phase 3 (Python):
   python manage.py --execute --context='{"operation": "create", "agent_output": {...}}'
   - Validates structure
   - Creates directories + scripts/, references/, templates/
   - Writes SKILL.md and supporting files

6. Report to user:
   "Created skill 'auto-format' with 4 files"