From wicked-garden
Capability router that decides which tools, skills, and agents to use for a task. Discovers CLI tools in PATH alongside MCP servers, skills, and agents. Reduces cognitive load on the main agent by making tool selection decisions. Use when: planning work for unfamiliar domains, evaluating task scope before execution, discovering what integrations or MCP servers could help, CLI detection, which tools are installed, available CLIs, building task execution strategies, or when unsure which tools, skills, or agents to use for current work.
npx claudepluginhub mikeparcewski/wicked-garden --plugin wicked-gardenThis skill uses the workspace's default tool permissions.
**Purpose**: Scout available capabilities and decide what to use for a task. Return actionable recommendations so the main agent can execute without discovery overhead.
Provides Ktor server patterns for routing DSL, plugins (auth, CORS, serialization), Koin DI, WebSockets, services, and testApplication testing.
Conducts multi-source web research with firecrawl and exa MCPs: searches, scrapes pages, synthesizes cited reports. For deep dives, competitive analysis, tech evaluations, or due diligence.
Provides demand forecasting, safety stock optimization, replenishment planning, and promotional lift estimation for multi-location retailers managing 300-800 SKUs.
Purpose: Scout available capabilities and decide what to use for a task. Return actionable recommendations so the main agent can execute without discovery overhead.
Most agents execute with whatever tools are obvious. They don't:
This skill changes that. It scouts, evaluates, and recommends—so the main agent can focus on execution.
Before executing this task, let me check what capabilities could help.
**Task**: Review PR #123 for security issues
**Discovery**:
- MCP: github (PR context), atlassian (link to Jira)
- Skills: qe (test scenarios), platform (security checks)
- Agents: wicked-garden:platform:security-engineer
**Recommendation**: Use security-engineer agent for review, github MCP for PR data.
For complex tasks, spawn integration-discovery as a subagent:
Task tool:
subagent_type: "Explore"
prompt: "Scout available capabilities for: [task description].
Return specific recommendations on which MCP servers,
skills, and agents to use."
What does this task need?
Check these sources (see refs/discovery-sources.md for details):
| Source | How to Check | What You Get |
|---|---|---|
| MCP Servers | ListMcpResourcesTool | External integrations |
| CLI Tools | command -v {tool} | Installed binaries in PATH |
| Skills | Check Skill tool description | Methodology/expertise |
| Agents | Check Task tool agent list | Specialized workers |
| Built-in Tools | Known set | File ops, search, web |
For each task need, find matching capabilities:
Task: "Implement user authentication"
| Need | Capability | Recommendation |
|------|------------|----------------|
| Security review | wicked-garden:platform:security-engineer | Use for auth review |
| Test strategy | wicked-garden:crew:gate-adjudicator | Use for test planning |
| Store decisions | wicked-garden:mem | Store auth decisions |
| Track work | MCP:atlassian or native TaskCreate | Track implementation |
Output a clear recommendation the main agent can act on:
## Capability Recommendations for: [Task]
**Use these**:
1. **wicked-garden:platform:security-engineer** - Review for security issues
2. **MCP:atlassian** - Link work to Jira ticket AUTH-123
**Available but optional**:
- wicked-garden:crew:gate-adjudicator - If test planning needed
- wicked-garden:mem - Store decisions for future reference
**Not available** (consider installing):
- No analytics MCP found - can't check auth failure metrics
Quick mappings for common tasks (see refs/task-patterns.md for full list):
| Task Type | Likely Capabilities |
|---|---|
| Code review | engineering agents, github MCP |
| Security audit | platform agents, qe |
| Architecture design | engineering agents, wicked-garden:mem |
| Bug investigation | engineering agents, error tracking MCP |
| Feature planning | product agents, project management MCP |
| Documentation | engineering agents, confluence MCP |
When multiple options exist:
Always return structured recommendations:
## Capability Recommendations
**Task**: [one-line task description]
### Recommended
| Capability | Type | Use For |
|------------|------|---------|
| name | MCP/Skill/Agent | specific purpose |
### Optional
| Capability | Type | Use If |
|------------|------|--------|
| name | type | condition |
### Not Available
| Need | Suggestion |
|------|------------|
| what's missing | how to get it |