digital-stoic-org

edit-tool

0
0
# Install this skill:
npx skills add digital-stoic-org/agent-skills --skill "edit-tool"

Install specific skill from multi-skill repository

# Description

Orchestrates creation of Claude Code tools (skills, commands, agents, scripts). Use when user requests creating, updating, or improving any Claude Code extension mechanism. Triggers include "create/make/new skill/command/agent/script", "tool for X", "slash command", "sub-agent", file paths with /skills/, /commands/, /agents/. Auto-triages based on token budget, frequency, context needs, and execution type. Delegates to edit-skill, edit-command, or edit-agent after explaining decision rationale.

# SKILL.md


name: edit-tool
description: Orchestrates creation of Claude Code tools (skills, commands, agents, scripts). Use when user requests creating, updating, or improving any Claude Code extension mechanism. Triggers include "create/make/new skill/command/agent/script", "tool for X", "slash command", "sub-agent", file paths with /skills/, /commands/, /agents/. Auto-triages based on token budget, frequency, context needs, and execution type. Delegates to edit-skill, edit-command, or edit-agent after explaining decision rationale.


Edit Tool Orchestrator

Automatically triages tool creation requests to the appropriate mechanism (skill, command, agent, or script).

Triage Process

CRITICAL: Always follow these steps:

  1. Analyze request against decision tree
  2. Explain decision to user with rationale
  3. Invoke appropriate skill or provide guidance

Decision Tree

graph TD
    A[Tool Request] --> B{Deterministic<br/>shell only?}
    B -->|Yes| C{AI needed to<br/>decide when/how?}
    C -->|No| BASH[✅ Bash Script<br/>0 tokens]
    C -->|Yes| D{Token<br/>budget?}
    D -->|<500| WRAP[✅ Skill + scripts/<br/>AI wraps code]
    D -->|>500| CMD1[✅ Command<br/>+ @filepath]

    B -->|No| E{Token<br/>budget?}
    E -->|>2000| AGENT[✅ Sub-Agent<br/>Isolated context]
    E -->|500-2000| F{Invocation?}
    F -->|User| CMD2[✅ Command<br/>User controls]
    F -->|Auto| AGENT

    E -->|<500| G{Frequency?}
    G -->|5+ per session| H{Specific<br/>capability?}
    H -->|Yes| SKILL[✅ Skill<br/>Auto-invoked]
    H -->|No| CMD2
    G -->|1-2 per session| CMD2
    G -->|Rare <1/10| DIRECT[❌ Direct Request<br/>No tool needed]

    style BASH fill:#90EE90,stroke:#000,stroke-width:2px,color:#000
    style SKILL fill:#FFD700,stroke:#000,stroke-width:2px,color:#000
    style WRAP fill:#FFA500,stroke:#000,stroke-width:2px,color:#000
    style CMD1 fill:#87CEEB,stroke:#000,stroke-width:2px,color:#000
    style CMD2 fill:#87CEEB,stroke:#000,stroke-width:2px,color:#000
    style AGENT fill:#DDA0DD,stroke:#000,stroke-width:2px,color:#000
    style DIRECT fill:#FFB6C1,stroke:#000,stroke-width:2px,color:#000

Decision Factors Reference

When path requires multiple criteria, consult:

Factor Key Question Result
Token Budget <500 / 500-2000 / >2000? Skill / Command / Agent
Frequency 5+ / 1-2 / <1 per 10 sessions? Skill / Command / Direct
Context Main / Isolated? Command / Agent
Scripts AI wrapper needed? Skill+scripts/ / Bash
Invocation User / Auto? Command / Skill
Capability Specific / Workflow? Skill / Command

Explanation Template

Explain decision: ✅ [TYPE] because: token budget (~X → range), frequency (pattern), key factor (dimension)

Example: ✅ SKILL because: ~300 tokens → <500, 10+/session, auto-invoked capability

Delegation

After explanation: Skilledit-skill | Commandedit-command | Agentedit-agent | Bash → Direct guidance (scripts/, #!/bin/bash, chmod +x, no tool)

Parallelization Check

Before any write operations, verify safety:

Operation Type Guidance
✅ Read-only Parallelize freely
⚠️ Writes (independent files) Sequential OR Plan Mode first
⚠️ Writes (dependent changes) Must sequence carefully
❌ Destructive operations Plan Mode MANDATORY

Plan Mode triggers:
- Creating >3 files
- Modifying >5 files
- Architectural changes
- Breaking changes

Tool Comparison Quick Reference

Tool Token Cost When to Use Context
Bash Script 0 (executed) Deterministic shell ops None
Skill <500 ideal Auto-invoked, frequent Shared (pollution)
Command 500-2000 ok User-triggered workflows User-controlled
Agent Unlimited Complex exploration Isolated

Common Patterns

Pattern When Solution
Script Wrapper AI decides timing for existing scripts Skill + scripts/
Verbose Workflow >1000 tokens, manual trigger Slash command
Research Task Multi-file exploration, autonomous Sub-agent (check Task tool first)

See reference.md for edge cases, conversion guide, and extended examples.

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.