89 lines
2.6 KiB
Markdown
89 lines
2.6 KiB
Markdown
---
|
|
name: orchestrate
|
|
allowed-tools: [Task, TodoWrite]
|
|
description: "Orchestrate complete Fabric pattern workflows"
|
|
argument-hint: [user_prompt]
|
|
category: utility
|
|
complexity: intermediate
|
|
mcp-servers: []
|
|
---
|
|
|
|
## Usage
|
|
```
|
|
/orchestrate [user_prompt]
|
|
```
|
|
|
|
## Arguments
|
|
- `user_prompt` - Describe your complete workflow need
|
|
|
|
## Execution
|
|
|
|
This command orchestrates the complete pattern workflow:
|
|
|
|
1. **Get Pattern Suggestions**:
|
|
Use Task tool to call pattern-suggester with user prompt
|
|
|
|
2. **Parse Pattern Sequence**:
|
|
Extract the recommended pattern sequence from the response
|
|
|
|
3. **Execute Pattern Chain**:
|
|
For each pattern in the sequence:
|
|
- Use Task tool to call pattern-executor with the pattern name
|
|
- Pass the previous pattern's output as input to the next pattern
|
|
- Store each output for chaining
|
|
|
|
4. **Return Final Result**:
|
|
Return the last pattern's output directly to the user
|
|
|
|
## Detailed Workflow
|
|
|
|
### Step 1: Get Suggestions
|
|
Call pattern-suggester agent with user prompt to get pattern sequence
|
|
|
|
### Step 2: Parse Response
|
|
Extract the pattern sequence from suggester's response (e.g., `analyze → extract → create_summary`)
|
|
|
|
### Step 3: Execute Each Pattern
|
|
For each pattern in the sequence, use Task tool with:
|
|
- description: "Execute [pattern_name] pattern"
|
|
- prompt: "Execute [pattern_name] pattern with this input: [previous_output or original_input]"
|
|
- subagent_type: "pattern-executor"
|
|
|
|
### Step 4: Chain Outputs
|
|
- First pattern receives the original user input
|
|
- Each subsequent pattern receives the output from the previous pattern
|
|
- Continue until all patterns are executed
|
|
|
|
### Step 5: Return Result
|
|
Return the final pattern's output directly without modification
|
|
|
|
## Examples
|
|
|
|
### Documentation Generation
|
|
```
|
|
/orchestrate "Document my codebase with clean, formatted output"
|
|
```
|
|
Workflow:
|
|
1. pattern-suggester returns: `analyze_code → extract_structure → create_documentation`
|
|
2. Execute analyze_code with original code
|
|
3. Execute extract_structure with code analysis
|
|
4. Execute create_documentation with structure data
|
|
5. Return final documentation
|
|
|
|
### Security Analysis
|
|
```
|
|
/orchestrate "Analyze security vulnerabilities and create a report"
|
|
```
|
|
Workflow:
|
|
1. pattern-suggester returns: `analyze_security → extract_vulnerabilities → create_report`
|
|
2. Execute analyze_security with codebase
|
|
3. Execute extract_vulnerabilities with security analysis
|
|
4. Execute create_report with vulnerability list
|
|
5. Return security report
|
|
|
|
## Important Notes
|
|
- The command handles all orchestration directly
|
|
- Each pattern's output becomes the next pattern's input
|
|
- No intermediate agents are involved
|
|
- Final output is returned unmodified
|