Initial commit

commands/exec.md

@@ -0,0 +1,43 @@
+---
+description: Execute a specific Fabric pattern by name
+argument-hint: [pattern_name] [user_prompt]
+allowed-tools: [Task]
+category: utility
+complexity: basic
+mcp-servers: []
+---
+
+## Usage
+```
+/exec [pattern_name] [user_prompt]
+```
+
+## Arguments
+- `pattern_name` - The name of the pattern to execute (e.g., "review_code", "summarize")
+- `user_prompt` - The input text to process
+
+## Examples
+```
+/exec review_code "analyze the login function"
+/exec summarize "last 5 commits"
+/exec analyze_security "[code here]"
+```
+
+## Execution
+
+This command delegates to the pattern-executor subagent which uses the Sonnet model for high-quality analysis.
+
+Use the pattern-executor subagent to execute the pattern with the following input:
+
+--- PATTERN NAME ---
+$1
+
+--- INPUT START ---
+$2
+--- INPUT END ---
+
+The pattern-executor will:
+1. Extract the specified pattern from `${CLAUDE_PLUGIN_ROOT}/.fabric-core/pattern_extracts.json`
+2. Apply the pattern to the provided input
+3. Generate comprehensive analysis using Sonnet model
+4. Return the formatted result

commands/orchestrate.md

@@ -0,0 +1,88 @@
+---
+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

commands/suggest.md

@@ -0,0 +1,37 @@
+---
+name: suggest
+argument-hint: [user_prompt]
+allowed-tools: [Task]
+description: "Suggest Fabric patterns based on user prompt analysis"
+category: utility
+complexity: basic
+mcp-servers: []
+---
+
+## Usage
+```
+/suggest [user_prompt]
+```
+
+## Arguments
+- `user_prompt` - Describe what you want to do
+
+## Execution
+
+Delegate to the `pattern-suggester` subagent with the provided user prompt for intelligent pattern suggestions based on semantic analysis.
+
+The pattern-suggester agent will:
+- Load and analyze the Fabric pattern library from `${CLAUDE_PLUGIN_ROOT}/.fabric-core/pattern_descriptions.json`
+- Perform deep semantic analysis of the user prompt
+- Identify primary intent, domain context, and specific requirements
+- Match patterns based on tags, semantic similarity, and use case alignment
+- Generate 3-5 targeted pattern recommendations with clear reasoning
+- Suggest single patterns for simple tasks or multi-pattern workflows for complex tasks
+- Provide alternative approaches for different outcomes
+
+## Example
+```
+/suggest "I need to analyze security vulnerabilities in my codebase"
+```
+
+This will invoke the pattern-suggester subagent to recommend security-focused analysis patterns from the Fabric library.