Devpipe MCP Server

# Devpipe MCP Server - Usage Examples This document provides detailed examples of how to use the devpipe MCP server with AI assistants. ## Example 1: Quick Pipeline Run **User:** "Run my devpipe pipeline with fast mode" **AI Assistant uses:** - Tool: `run_pipeline` - Parameters: `{ "fast": true }` **Result:** Pipeline executes skipping slow tasks, returns execution summary with task results. --- ## Example 2: Debugging a Failed Task **User:** "The lint task failed in my last run. Can you help me figure out why?" **AI Assistant workflow:** 1. Uses `get_last_run` to get run metadata 2. Uses `view_run_logs` with `{ "taskId": "lint" }` to get the log 3. Analyzes the error and suggests fixes **Result:** Detailed analysis of the failure with actionable recommendations. --- ## Example 3: Configuration Analysis **User:** "Analyze my pipeline configuration and suggest improvements" **AI Assistant uses:** - Prompt: `analyze-config` - Resource: `devpipe://config` - Tool: `list_tasks` **Result:** Comprehensive analysis covering: - Task organization and phases - Parallel execution opportunities - Missing common checks - Configuration best practices --- ## Example 4: Creating a New Task **User:** "Help me create a task to run Go tests with coverage" **AI Assistant uses:** - Prompt: `create-task` with `{ "technology": "Go", "taskType": "test" }` **Result:** Complete TOML configuration: ```toml [tasks.go-test] name = "Go Tests with Coverage" desc = "Run Go unit tests and generate coverage report" command = "go test -v -race -coverprofile=coverage.out ./..." type = "test" workdir = "." outputType = "junit" outputPath = "test-results.xml" ``` --- ## Example 5: Running Specific Tasks **User:** "Run only the lint and format tasks, skip everything else" **AI Assistant uses:** - Tool: `run_pipeline` - Parameters: `{ "only": ["lint", "format"] }` **Result:** Only specified tasks execute. --- ## Example 6: Validating Configuration **User:** "I just modified my config.toml. Can you validate it?" **AI Assistant uses:** - Tool: `validate_config` - Parameters: `{ "configs": ["config.toml"] }` **Result:** Validation results showing any syntax or structural errors. --- ## Example 7: Security Review **User:** "Review the security findings from my last SARIF scan" **AI Assistant uses:** - Prompt: `security-review` - Tool: `get_dashboard_data` - Resource: `devpipe://summary` **Result:** Security analysis with: - Findings grouped by severity - Critical issues highlighted - Remediation recommendations - Suggested additional security checks --- ## Example 8: Pipeline Optimization **User:** "My pipeline is taking too long. How can I speed it up?" **AI Assistant uses:** - Prompt: `optimize-pipeline` - Tool: `list_tasks` - Tool: `get_dashboard_data` **Result:** Optimization suggestions: - Tasks that can run in parallel - Fast mode configuration - Caching strategies - Git integration for change-based runs --- ## Example 9: Dry Run Before Commit **User:** "Show me what would run if I execute devpipe right now" **AI Assistant uses:** - Tool: `run_pipeline` - Parameters: `{ "dryRun": true, "verbose": true }` **Result:** List of tasks that would execute without actually running them. --- ## Example 10: Git-Based Execution **User:** "Run checks only on files I've changed since main branch" **AI Assistant uses:** - Tool: `run_pipeline` - Parameters: `{ "since": "origin/main" }` **Result:** Pipeline runs only on changed files, saving time. --- ## Example 11: Viewing All Resources **User:** "Show me the current state of my pipeline" **AI Assistant accesses:** - Resource: `devpipe://config` - Current configuration - Resource: `devpipe://tasks` - All task definitions - Resource: `devpipe://last-run` - Latest run results - Resource: `devpipe://summary` - Overall summary **Result:** Complete overview of pipeline state and history. --- ## Example 12: Multi-Config Validation **User:** "Validate all my config files" **AI Assistant uses:** - Tool: `validate_config` - Parameters: `{ "configs": ["config.toml", "config.ci.toml", "config.prod.toml"] }` **Result:** Validation results for all configuration files. --- ## Example 13: Task Log Investigation **User:** "Show me what happened during the build task" **AI Assistant uses:** - Tool: `view_run_logs` - Parameters: `{ "taskId": "build" }` **Result:** Complete build task log output. --- ## Example 14: Metrics Parsing **User:** "Parse the JUnit test results from my last run" **AI Assistant uses:** - Tool: `get_last_run` to find the run directory - Tool: `parse_metrics` with `{ "metricsPath": ".devpipe/runs/.../junit.xml", "format": "junit" }` **Result:** Parsed test metrics with pass/fail counts and details. --- ## Example 15: Installation Check **User:** "Is devpipe installed on this system?" **AI Assistant uses:** - Tool: `check_devpipe` **Result:** Installation status and version information, or installation instructions if not found. --- ## Complex Workflow Example **User:** "I want to set up a comprehensive Go project pipeline" **AI Assistant workflow:** 1. Uses `check_devpipe` to verify installation 2. Uses `list_tasks` to see current configuration 3. Uses `create-task` prompt multiple times for: - Go formatting (gofmt) - Go linting (golangci-lint) - Go tests with coverage - Go build - Security scanning (gosec) 4. Suggests phase organization for parallel execution 5. Recommends metrics configuration (JUnit for tests, SARIF for security) 6. Uses `validate_config` to verify the new configuration 7. Uses `run_pipeline` with `{ "dryRun": true }` to preview execution **Result:** Complete, validated pipeline configuration ready to use. --- ## Tips for AI Assistants When using this MCP server: 1. **Always check devpipe installation first** for new users 2. **Use dry-run mode** before making destructive changes 3. **Combine tools** for comprehensive analysis (e.g., get_last_run + view_run_logs) 4. **Leverage prompts** for complex workflows (analyze-config, optimize-pipeline) 5. **Access resources** for context before making suggestions 6. **Parse metrics** to provide data-driven recommendations 7. **Validate configs** after suggesting changes ## Common Patterns ### Pattern: Debug Workflow 1. `get_last_run` - Get run metadata 2. `view_run_logs` - Get specific task log 3. Analyze and suggest fixes ### Pattern: Optimization Workflow 1. `list_tasks` - See current tasks 2. `get_dashboard_data` - Get performance data 3. `optimize-pipeline` prompt - Get suggestions 4. `validate_config` - Verify changes ### Pattern: Setup Workflow 1. `check_devpipe` - Verify installation 2. `create-task` prompts - Generate tasks 3. `validate_config` - Verify configuration 4. `run_pipeline` with dryRun - Preview execution