Skip to main content
Glama
drewkhoury

Devpipe MCP Server

by drewkhoury
OverviewInspectSchemaRelated ServersScoreDiscussions
TypeScript
Local

Devpipe MCP Server

A Model Context Protocol (MCP) server that enables AI assistants to interact with devpipe - a fast, local pipeline runner for development workflows.

Features

This MCP server provides AI assistants with the ability to:

  • šŸ“‹ List and analyze tasks from devpipe configurations (with verbose stats)

  • šŸš€ Run pipelines with full control over execution flags

  • āœ… Validate configurations before running

  • šŸ“Š Access run results and metrics (JUnit, SARIF)

  • šŸ” Debug failures by reading task logs

  • šŸ’” Suggest optimizations for pipeline configurations

  • šŸ›”ļø Review security findings from SARIF reports

  • šŸ”§ Auto-detect technologies and suggest missing tasks

  • āš” Generate task configurations from templates

  • šŸ“ Create complete configs from scratch

  • šŸ”„ Generate CI/CD configs (GitHub Actions, GitLab CI)

Prerequisites

  • Node.js 18 or higher

  • devpipe v0.0.8 or higher installed and accessible in PATH

    brew install drewkhoury/tap/devpipe

    Note: This MCP is optimized for devpipe v0.0.8+ which includes updated default values and improved documentation.

Installation

Option 1: Install from npm (when published)

npm install -g devpipe-mcp

Option 2: Install from source

git clone https://github.com/drewkhoury/devpipe-mcp.git cd devpipe-mcp npm install npm run build npm link

Configuration

For Windsurf/Cascade

Add to your Windsurf MCP settings file (usually ~/.windsurf/mcp.json or similar):

{ "mcpServers": { "devpipe": { "command": "devpipe-mcp" } } }

For Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{ "mcpServers": { "devpipe": { "command": "devpipe-mcp" } } }

For Other MCP Clients

The server runs on stdio, so you can connect any MCP client using:

devpipe-mcp

Usage

Once configured, you can interact with devpipe through your AI assistant. Here are some example requests:

List Tasks

"Show me all the tasks in my devpipe configuration" "What tasks are defined in config.toml?"

Run Pipeline

"Run the devpipe pipeline" "Run only the lint and test tasks" "Run devpipe with --fast and --fail-fast flags" "Execute devpipe in dry-run mode"

Validate Configuration

"Validate my devpipe config" "Check if config.toml is valid"

Debug Failures

"Why did the lint task fail?" "Show me the logs for the build task" "What went wrong in the last run?"

Analyze and Optimize

"Analyze my pipeline configuration" "Suggest optimizations for my devpipe setup" "How can I make my pipeline faster?"

Create Tasks

"Create a devpipe task for running Go tests" "Help me add a Python linting task" "Generate a task configuration for ESLint"

Bootstrap New Projects

"Create a devpipe config for this project" "Analyze my project and suggest tasks" "What technologies did you detect in /path/to/project?"

Generate CI/CD

"Generate a GitHub Actions workflow for devpipe" "Create a GitLab CI config from my devpipe setup"

Security Review

"Review the security findings from my last run" "What security issues were found?" "Analyze SARIF results"

MCP Tools

The server provides the following tools:

list_tasks

Parse and list all tasks from a config.toml file.

Parameters:

  • config (optional): Path to config.toml file

Example:

{ "config": "./config.toml" }

run_pipeline

Execute devpipe with specified flags.

Parameters:

  • config (optional): Path to config.toml

  • only (optional): Array of task IDs to run

  • skip (optional): Array of task IDs to skip

  • since (optional): Git reference for change-based runs

  • fixType (optional): auto, helper, or none

  • ui (optional): basic or full

  • dashboard (optional): Show dashboard view

  • failFast (optional): Stop on first failure

  • fast (optional): Skip slow tasks

  • dryRun (optional): Show what would run

  • verbose (optional): Verbose output

  • noColor (optional): Disable colors

Example:

{ "only": ["lint", "test"], "fast": true, "failFast": true }

validate_config

Validate devpipe configuration files.

Parameters:

  • configs (optional): Array of config file paths

Example:

{ "configs": ["config.toml", "config.prod.toml"] }

get_last_run

Get results from the most recent pipeline run.

Parameters:

  • config (optional): Path to config.toml

view_run_logs

Read logs from a specific task or the entire pipeline.

Parameters:

  • taskId (optional): Task ID to view logs for

  • config (optional): Path to config.toml

Example:

{ "taskId": "lint" }

parse_metrics

Parse JUnit or SARIF metrics files.

Parameters:

  • metricsPath (required): Path to metrics file

  • format (required): junit or sarif

Example:

{ "metricsPath": ".devpipe/runs/latest/metrics.sarif", "format": "sarif" }

get_dashboard_data

Extract aggregated data from summary.json.

Parameters:

  • config (optional): Path to config.toml

check_devpipe

Check if devpipe is installed and get version info.

list_tasks_verbose

List tasks using devpipe list --verbose command with execution statistics.

Parameters:

  • config (optional): Path to config.toml file

Example:

{ "config": "./config.toml" }

Output: Shows task table with average execution times and statistics.

analyze_project

Analyze project directory to detect technologies and suggest missing tasks.

Parameters:

  • projectPath (optional): Path to project directory (defaults to current)

Example:

{ "projectPath": "/path/to/project" }

Output:

{ "projectPath": "/path/to/project", "detectedTechnologies": ["Go", "Docker"], "suggestedTasks": [ { "technology": "Go", "taskType": "check-format", "reason": "go fmt for formatting" }, { "technology": "Go", "taskType": "check-lint", "reason": "golangci-lint for linting" } ], "summary": "Found 2 technologies with 5 suggested tasks" }

generate_task

Generate task configuration from template for a specific technology or phase header.

Parameters:

  • technology (required): Technology name (e.g., "Go", "Python", "Node.js", "TypeScript", "Rust") or "phase" for phase headers

  • taskType (required): Task type (e.g., "check-format", "check-lint", "test-unit", "build") or phase name

  • taskId (optional): Custom task ID for regular tasks, or description for phase headers

Example (Regular Task):

{ "technology": "Go", "taskType": "check-lint", "taskId": "golangci-lint" }

Output:

[tasks.golangci-lint] name = "Golang CI Lint" desc = "Runs comprehensive linting on Go code" type = "check" command = "golangci-lint run" fixType = "auto" fixCommand = "golangci-lint run --fix"

Example (Phase Header):

{ "technology": "phase", "taskType": "Validation", "taskId": "Static analysis and tests" }

Output:

[tasks.phase-validation] name = "Validation" desc = "Static analysis and tests"

Note: Phase headers have no required fields - they're organizational markers. Common practice is to include name or desc (or both), but neither is strictly required.

Supported Technologies:

  • Go: check-format, check-lint, check-static, test-unit, build

  • Python: check-format, check-lint, check-types, test-unit

  • Node.js: check-lint, test-unit, build

  • TypeScript: check-types

  • phase: Creates phase headers (organizational markers, no command/type)

create_config

Create a complete config.toml file from scratch with auto-detected tasks.

Parameters:

  • projectPath (optional): Path to project directory (defaults to current)

  • includeDefaults (optional): Include [defaults] section (default: true)

  • autoDetect (optional): Auto-detect technologies and generate tasks (default: true)

Example:

{ "projectPath": "/path/to/project", "includeDefaults": true, "autoDetect": true }

Output: Complete config.toml with:

  • Defaults section (outputRoot, fastThreshold=300s, animationRefreshMs=500ms, git settings)

  • Task defaults (enabled, workdir)

  • Auto-detected tasks organized by phase

  • Ready-to-use TOML configuration compatible with devpipe v0.0.8+

Use Case: Bootstrap a new project with devpipe configuration.

Note: Generated configs use devpipe v0.0.8 defaults (fastThreshold=300s, not 5000ms).

generate_ci_config

Generate CI/CD configuration file (GitHub Actions or GitLab CI) from devpipe config.

Parameters:

  • config (optional): Path to config.toml file

  • platform (required): github or gitlab

Example:

{ "config": "./config.toml", "platform": "github" }

Output (GitHub Actions):

name: CI Pipeline on: push: branches: [ main, develop ] pull_request: branches: [ main ] jobs: devpipe: runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 - name: Install devpipe run: | curl -L https://github.com/drewkhoury/devpipe/releases/latest/download/devpipe-linux-amd64 -o devpipe chmod +x devpipe sudo mv devpipe /usr/local/bin/ - name: Run devpipe run: devpipe --fail-fast - name: Upload results if: always() uses: actions/upload-artifact@v4 with: name: devpipe-results path: .devpipe/

MCP Resources

The server exposes these resources:

  • devpipe://config - Current config.toml contents

  • devpipe://tasks - All task definitions

  • devpipe://last-run - Most recent run results

  • devpipe://summary - Aggregated pipeline summary

  • devpipe://schema - JSON Schema for config.toml validation (fetched from official devpipe repo)

MCP Prompts

Pre-configured prompts for common workflows:

analyze-config

Analyze the devpipe configuration and suggest improvements.

debug-failure

Help debug why a specific task failed.

Arguments:

  • taskId (required): The task that failed

optimize-pipeline

Suggest optimizations for the pipeline.

create-task

Help create a new task for a technology.

Arguments:

  • technology (required): Technology name (e.g., "Go", "Python")

  • taskType (optional): check, build, or test

security-review

Review SARIF security findings and provide recommendations.

Examples

See EXAMPLES.md for detailed usage examples and workflows.

Quick Example

User: "Run my devpipe pipeline with fast mode" Assistant: *Uses run_pipeline tool with { "fast": true }* Result: Pipeline executes, skipping slow tasks

Development

Building from Source

git clone https://github.com/drewkhoury/devpipe-mcp.git cd devpipe-mcp npm install npm run build

Project Structure

devpipe-mcp/ ā”œā”€ā”€ src/ ā”‚ ā”œā”€ā”€ index.ts # Main MCP server ā”‚ ā”œā”€ā”€ types.ts # Type definitions ā”‚ ā””ā”€ā”€ utils.ts # Utility functions ā”œā”€ā”€ examples/ # Example configs ā”œā”€ā”€ dist/ # Compiled output ā””ā”€ā”€ README.md

Watch Mode

npm run watch

Troubleshooting

devpipe not found

If you get "devpipe not found" errors:

# Install devpipe brew install drewkhoury/tap/devpipe # Verify installation devpipe --version

Config file not found

The MCP server searches for config.toml in:

  1. Current directory

  2. Parent directories (up to root)

You can also specify the config path explicitly in tool calls.

Permission errors

Ensure the MCP server has permission to:

  • Read config files

  • Execute devpipe commands

  • Access the .devpipe output directory

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE file for details.

Related Projects

Support

Changelog

See CHANGELOG.md for version history and changes.

One-click Deploy
A
security ā€“ no known vulnerabilities
A
license - permissive license
A
quality - confirmed to work

How are these scores calculated?

Resources

Tools

View all tools

New MCP Servers

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/drewkhoury/devpipe-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server