// Architecture and design planning for new Claude Code skills. Guides through use case definition, complexity tier selection, sub-skill decomposition, and file structure planning. Use when user says "plan skill", "design skill", "skill architecture", or "skill planning".
Benchmark Claude Code skill performance with variance analysis, tracking pass rate, execution time, and token usage across iterations. Runs multiple trials per eval for statistical reliability, aggregates results into benchmark.json, and generates comparison reports between skill versions. Use when user says "benchmark skill", "measure skill performance", "skill metrics", "compare skill versions", "skill performance", "track skill improvement", "skill regression test", or "skill A/B test".
Run evaluation pipelines on Claude Code skills to test triggering accuracy, workflow correctness, and output quality. Spawns executor, grader, comparator, and analyzer sub-agents for parallel evaluation. Generates eval_metadata.json, grading.json, and feedback reports. Use when user says "eval skill", "test skill", "run evals", "evaluate skill", "skill evals", "test skill quality", "run skill tests", or "skill evaluation".
Ultimate Claude Code skill creator and architect. Designs, scaffolds, builds, reviews, evolves, and publishes production-grade Claude Code skills following the Agent Skills open standard and 3-layer architecture (directive, orchestration, execution). Handles single-file skills, multi-skill orchestrators with sub-skills and subagents, MCP-enhanced workflows, and full skill ecosystems. Industry detection for skill domain. Triggers on: "create skill", "build skill", "new skill", "skill creator", "skill builder", "skill-forge", "design skill", "scaffold skill", "review skill", "improve skill", "publish skill", "skill architecture", "convert skill", "port skill", "multi-platform", "cross-platform", "eval skill", "test skill", "benchmark skill", "skill evals", "measure skill", "skill performance", "skill A/B test".
Improve and iterate on existing Claude Code skills based on usage feedback, test results, or changing requirements. Handles under/over-triggering fixes, instruction refinement, new sub-skill addition, and architecture evolution. Use when user says "improve skill", "fix skill", "skill not triggering", "skill triggers too much", "update skill", or "evolve skill".
Audit and validate existing Claude Code skills for quality, triggering accuracy, structure compliance, and best practices. Scores skills on a 0-100 scale and provides prioritized improvement recommendations. Use when user says "review skill", "audit skill", "check skill", "validate skill", or "skill quality".
Scaffold and build Claude Code skills from plans or descriptions. Generates SKILL.md files, sub-skills, scripts, references, agents, and templates following the Agent Skills standard. Use when user says "build skill", "scaffold skill", "generate skill", "create SKILL.md", or "implement skill".
| name | skill-forge-plan |
| description | Architecture and design planning for new Claude Code skills. Guides through use case definition, complexity tier selection, sub-skill decomposition, and file structure planning. Use when user says "plan skill", "design skill", "skill architecture", or "skill planning". |
Ask the user these questions (adapt based on context):
For each use case, define:
Use Case: [Name]
Trigger: User says "[phrases]"
Steps:
1. [First action]
2. [Decision point or validation]
3. [Next action]
Result: [What success looks like]
Tools Needed: [built-in, MCP, scripts]
Evaluate based on answers:
| Signal | Tier 1 | Tier 2 | Tier 3 | Tier 4 |
|---|---|---|---|---|
| Use cases | 1-2 | 2-3 | 4-8 | 8+ |
| Needs scripts? | No | Yes | Maybe | Yes |
| Sub-skills needed? | No | No | Yes | Yes |
| Parallel execution? | No | No | No | Yes |
| Reference docs? | No | Maybe | Yes | Yes |
| Industry templates? | No | No | Maybe | Yes |
Decision matrix:
Based on tier, generate the architecture:
Tier 1 Output:
skill-name/
SKILL.md
Tier 2 Output:
skill-name/
SKILL.md
scripts/
validate.py
process.py
references/
domain-knowledge.md
Tier 3 Output:
skill-name/ # Main orchestrator
SKILL.md
references/
shared-reference.md
skills/
skill-name-sub1/
SKILL.md
skill-name-sub2/
SKILL.md
Tier 4 Output:
skill-name/ # Main orchestrator
SKILL.md
references/
ref1.md
ref2.md
scripts/
script1.py
script2.py
assets/
template1.md
template2.md
skills/
skill-name-sub1/
SKILL.md
skill-name-sub2/
SKILL.md
...
agents/
skill-name-role1.md
skill-name-role2.md
For each sub-skill, define:
{parent}-{function} (kebab-case)Design the command routing:
| Command | Routes to | Purpose |
|---------|-----------|---------|
| /skill-name | main SKILL.md | Interactive mode |
| /skill-name sub1 | skills/skill-name-sub1/ | Sub-workflow 1 |
| /skill-name sub2 | skills/skill-name-sub2/ | Sub-workflow 2 |
Identify knowledge that should be extracted to reference files:
Rule of thumb: If information is >50 lines and only needed for specific sub-workflows,
extract it to references/.
Create a structured plan document:
# Skill Plan: [name]
## Overview
- Domain: [domain]
- Tier: [1-4]
- Sub-skills: [count]
- Scripts: [count]
## Use Cases
[list from Step 2]
## Architecture
[diagram from Step 4]
## Sub-Skills
[details from Step 5]
## Routing
[table from Step 6]
## Reference Files
[list from Step 7]
## Next Steps
Run `/skill-forge build [name]` to scaffold the skill.
User: "I want to create a skill for managing Docker containers and Kubernetes deployments"
Discovery reveals:
Assessment: Tier 3 (multi-skill orchestrator)
Architecture:
devops/ # Main orchestrator
SKILL.md
scripts/
health_check.py # Cluster health check
log_parser.py # Log analysis
references/
k8s-patterns.md # Deployment patterns
docker-best-practices.md
skills/
devops-docker/SKILL.md # Container management
devops-k8s/SKILL.md # Kubernetes deployments
devops-monitor/SKILL.md # Monitoring and alerts
devops-logs/SKILL.md # Log analysis
devops-scale/SKILL.md # Scaling strategies
devops-fix/SKILL.md # Troubleshooting