agents-md-generator 0.4.1

MCP server that analyzes codebases and generates AGENTS.md files

agents-md-generator

MCP server that analyzes codebases with tree-sitter and generates AGENTS.md files.

Compatible with any MCP-capable client: Claude Code, Gemini CLI, Cursor, Windsurf, and others.

How it works: The server does all the heavy lifting locally — AST parsing, incremental change detection, environment variable scanning, entry point detection. It writes a compact structured payload to disk and returns step-by-step instructions to your AI client. The client reads the payload and writes AGENTS.md. No large data travels over the MCP wire.

Supported Languages

Python · C# · TypeScript · JavaScript · Go

---

Installation

See INSTALLATION.md for the full guide including prerequisites and troubleshooting.

Requirements: Python 3.11+, uv, Git, and any MCP-compatible client.

Claude Code

claude mcp add agents-md uvx agents-md-generator

Or add it manually to ~/.claude.json (Linux/macOS) or %USERPROFILE%\.claude.json (Windows):

{
  "mcpServers": {
    "agents-md": {
      "command": "uvx",
      "args": ["agents-md-generator"]
    }
  }
}

Gemini CLI

Add it to ~/.gemini/settings.json:

{
  "mcpServers": {
    "agents-md": {
      "command": "uvx",
      "args": ["agents-md-generator"]
    }
  }
}

Other MCP clients (Cursor, Windsurf, etc.)

The server uses stdio transport. Add this entry to your client's MCP config under mcpServers:

"agents-md": {
  "command": "uvx",
  "args": ["agents-md-generator"]
}

Restart your client — uvx downloads the package automatically on first run.

---

Usage

Once registered, ask your AI client:

"Generate the AGENTS.md for this project"

The client will call scan_codebase automatically.

Tool Parameters

Parameter	Type	Default	Description
project_path	string	"."	Path to the project root
force_full_scan	boolean	false	Ignore cache and rescan everything from scratch

Note on force_full_scan: Use this only when explicitly requested. When asking Claude to improve or update an existing AGENTS.md, leave it as false — the incremental scan already provides all the data needed.

---

What Gets Generated

The generated AGENTS.md follows the agents.md open standard. It is written as a README for AI agents, not as documentation for humans. Sections include:

• Project Overview — tech stack and top-level architecture shape
• Architecture & Data Flow — detected layers or domains with data flow direction
• Conventions & Patterns — naming rules, export contracts, import rules, and how to add new entities end-to-end
• Environment Variables — variables detected in source files and .env.example
• Setup Commands — exact install and run commands from package.json, Makefile, etc.
• Development Workflow — build, watch, and dev server commands
• Testing Instructions — test commands and framework info (if detected)
• Code Style — lint/format commands (if config files detected)
• Build and Deployment — CI pipeline info (if detected)

Sections with no detected data are omitted entirely.

---

How Incremental Scanning Works

1. First run (cold start): All git-tracked source files are parsed with tree-sitter and cached
2. Subsequent runs: Only files whose SHA-256 hash changed since the last scan are re-parsed
3. Semantic diff: For modified files, only changed public symbols are included in the payload
4. No source changes? The tool stops and asks whether you want to improve the existing AGENTS.md content anyway
5. Private symbols and test file internals are excluded from both cache and payload — only the public API surface matters for AGENTS.md

How Large Payloads Are Streamed

For large codebases the analysis payload can be too big to return inline over the MCP wire. The server handles this transparently through a second tool: read_payload_chunk.

Flow:

1. scan_codebase runs the full analysis, writes the payload to disk, and returns a small response with total_chunks and instructions
2. The client calls read_payload_chunk(project_path, chunk_index=0), then increments chunk_index until the response contains has_more: false
3. The client concatenates all data fields in order and parses the result as JSON
4. The payload file is automatically deleted after the last chunk is read

This flow is pure MCP — no filesystem access required from the client side. Any MCP-compatible client can follow it.

Cache and Payload Location

All runtime artifacts are stored outside your project, in the user cache directory:

~/.cache/agents-md-generator/<project-hash>/cache.json  ← incremental scan cache

The <project-hash> is a SHA-256 of the project's absolute path — unique per project. Nothing is written to your repository.

Note: The server also writes a temporary payload.json to this directory during analysis, but it is managed entirely by the read_payload_chunk tool and deleted automatically after the last chunk is read. You never need to access it directly.

---

Project Configuration

Create .agents-config.json at your project root to customize behavior. This file is optional — all fields have defaults.

{
  "project_size": "medium",
  "exclude": [
    "**/node_modules/**",
    "**/bin/**",
    "**/obj/**",
    "**/.git/**",
    "**/dist/**",
    "**/build/**",
    "**/__pycache__/**",
    "**/*.min.js",
    "**/*.min.css",
    "**/*.bundle.js",
    "**/vendor/**",
    "**/packages/**",
    "**/.venv/**",
    "**/venv/**",
    "**/bower_components/**",
    "**/app/lib/**",
    "**/wwwroot/lib/**",
    "**/wwwroot/libs/**",
    "**/static/vendor/**",
    "**/public/vendor/**",
    "**/assets/vendor/**",
    "**/site-packages/**"
  ],
  "include": [],
  "languages": "auto",
  "agents_md_path": "./AGENTS.md",
  "max_file_size_bytes": 1048576
}

Options

Key	Default	Description
project_size	"medium"	Project scale — tunes all internal caps and thresholds (see Project Size Profiles)
exclude	(see above)	Glob patterns to exclude from analysis
include	[]	If non-empty, only analyze files matching these patterns
languages	"auto"	"auto" detects all supported languages, or pass a list like ["typescript", "python"]
agents_md_path	"./AGENTS.md"	Output path for the generated file
max_file_size_bytes	1048576	Files larger than this are skipped (default: 1 MB)

You can commit .agents-config.json to share settings with your team.

Project Size Profiles

The project_size setting controls how aggressively the payload is compressed. A single knob tunes all internal caps — methods per class, symbols per file, directory aggregation, route caps, tree depth, and impact filtering.

Profile	Lines (guidance)	Impact filter	Description
"small"	0–15k	medium	Generous caps — nearly everything is included. Best for small projects where full visibility matters.
"medium" (default)	15k–50k	medium	Balanced caps suitable for most projects.
"large"	50k+	high	Aggressive compression — only structural/breaking changes in diffs, more directory collapsing, tighter symbol caps.

Detailed profile values:

Constant	Small	Medium	Large
Methods per class	30	12	8
Symbols per file	40	20	10
Dir aggregation threshold	20	10	5
Files per layer (before overflow)	15	8	5
Aggregation sample size	5	4	3
Route controllers cap	30	15	10
Routes per controller	15	8	5
Go handlers cap	15	8	5
Directory tree depth	4	3	2
Impact filter	medium	medium	high

---

What the Analysis Detects

Environment Variables

The server scans all source files for environment variable references using language-specific patterns:

Language	Pattern detected
JavaScript / TypeScript	process.env.VAR_NAME
Python	os.environ['VAR'], os.getenv('VAR')
Go	os.Getenv("VAR")
Ruby	ENV['VAR']
Rust	env!("VAR"), var("VAR")

It also parses .env.example, .env.template, and .env.sample files at the project root.

Entry Points

Files named index, main, app, server, program, bootstrap, or startup (with any supported extension) are detected as entry points and annotated with their inferred role (e.g., "HTTP server bootstrap", "Electron main process").

Public API Surface

Tree-sitter parses each source file and extracts public symbols — classes, functions, methods, interfaces — filtering out private/protected members and underscore-prefixed symbols. For classes and structs, constructors (when they have parameters) and public properties are also included, revealing dependency injection patterns and data shapes. Interface methods are always included as they define the public contract. These are used to detect naming conventions, DI patterns, and export contracts across layers.

Architectural Distillation

For large codebases, the tool applies several heuristics to ensure the payload remains high-signal:

• Boilerplate Suppression: Common directories like Migrations, bin, obj, and Properties are automatically flagged and collapsed in the project structure, preventing them from bloating the directory listing.
• Low-Entropy Summarization: Files that primarily contain data structures (DTOs, Entities) with no logic methods are "minified". Instead of listing every property, the tool provides a high-level summary (e.g., "Contains 25 DTO classes").
• Semantic Clustering: The aggregator groups these minified summaries at the directory level, allowing the consuming AI to understand entire data layers through a single line of signal.
• Instruction Prioritization: Foundation mandates (instructions) are placed at the very top of the payload, ensuring the AI agent understands the project's "Rules of Engagement" before processing the code architecture.

---

Credits

AGENTS.md format based on the open agents.md standard.