Agent Install
https://github.com/millionco/agent-install- Category
- AI Agents
- Rank
- No. 1483Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- millionco
- GitHub
- 53 stars
- Date
About
Install agent skills and MCPs with one API call. Simple installer for the modern agent stack.
What it does
Agent Install translates a requested skill, server, or guidance change into native configuration for many coding agents. It accepts local directories, git repositories, direct web sources, packages, remote endpoints, and commands, then targets selected agents through either a Node library or command line.
Why it's ranked here
This is a strong compatibility layer because it handles several configuration formats and agent-specific transformations behind one consistent interface. Its broad source parsing, project and global targets, comment-preserving JSONC writes, and matching library and command-line verbs make it practical for automation.
What's good
Skill sources cover local content, major git hosts, SSH remotes, direct documents, and well-known discovery endpoints. Repository clones are shallow, time-limited, and cleaned up defensively. MCP support accounts for transport restrictions, platform-specific locations, configuration keys, and transformations required by individual agents.
Tradeoffs
The package requires Node 22 or newer, while repository-based skill installation also depends on Git. MCP capabilities vary by agent: some lack project scope, and Claude Desktop accepts only standard-input servers. Configuration is local and file-based, so this does not replace server hosting or skill authoring.
How to use it well
Use it in Node build scripts, continuous integration, bootstrap tools, or occasional command-line setup when one project must configure several coding agents consistently. Select target agents explicitly and choose project or global scope carefully. Bring a separate workflow for creating skills and operating the MCP servers themselves.
Technical notes+
packages/agent-install/vite.config.ts builds CommonJS, ESM, and declaration outputs from packages/agent-install/src/index.ts, packages/agent-install/src/skill.ts, packages/agent-install/src/mcp.ts, packages/agent-install/src/agents-md.ts, and packages/agent-install/src/cli.ts. The first four expose namespace and subpath library surfaces, while packages/agent-install/src/cli.ts registers Commander subcommands and parses asynchronously. packages/agent-install/bin/agent-install.mjs optionally enables Node's compile cache before loading the built CLI. packages/agent-install/src/mcp/agents.ts maps platform-specific global and project locations, formats, keys, transports, detection logic, and per-agent transforms. packages/agent-install/src/skill/git.ts performs depth-one clones without interactive prompts or Git LFS downloads, applies a configurable timeout, classifies authentication and timeout failures, and restricts cleanup to the system temporary directory.
Observed
- License
- MIT-licensed open-source software
- Primary language
- TypeScript
- Packaging
- npm package with pnpm workspace development
- Interfaces
- Node library and command-line interface
- Runtime
- Node.js 22 or newer
- Build outputs
- CommonJS, ECMAScript modules, and TypeScript declarations
- Configuration formats
- JSON, JSONC, YAML, and TOML
- Platform handling
- Separate configuration path logic for Windows, macOS, and other XDG-style systems
Read from README.md, package.json, packages/agent-install/vite.config.ts, packages/agent-install/src/cli.ts, packages/agent-install/src/mcp.ts, packages/agent-install/src/index.ts, packages/agent-install/src/skill.ts, packages/agent-install/src/agents-md.ts, packages/agent-install/bin/agent-install.mjs, packages/agent-install/src/mcp/list.ts, packages/agent-install/src/mcp/index.ts, packages/agent-install/src/mcp/types.ts, packages/agent-install/src/skill/git.ts, packages/agent-install/src/mcp/agents.ts, packages/agent-install/src/mcp/remove.ts.
What it can do
Install agent skills via API
API call with skill specifications → Installed agent skills
Install MCPs (Model Control Protocols) via API
API call with MCP specifications → Installed MCP components
Deploy complete agent stack with single API call
API request with agent stack configuration → Fully deployed agent environment
Manage agent component dependencies
Component requirements and dependencies list → Resolved and installed dependencies
Configure agent runtime environment
Environment configuration parameters → Configured agent runtime
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.