fips-agents-cli 0.7.0

CLI tool for creating and managing FIPS-compliant AI agent projects

FIPS Agents CLI

A command-line tool for creating and managing FIPS-compliant AI agent projects. Scaffolds MCP (Model Context Protocol) servers, AI agent projects, and ModelCar containers from production-ready templates.

Features

• 🚀 Quick project scaffolding from templates
• 📦 MCP server, AI agent, Go gateway, chat UI, sandbox, and ModelCar project generation
• 🔧 Automatic project customization (pyproject.toml, module names, entry points)
• ⚡ Component generation (tools, resources, prompts, middleware) with Jinja2 templates
• 🎨 Beautiful CLI output with Rich
• ✅ Git repository initialization and GitHub integration
• 🧪 Comprehensive test coverage with auto-run on generated components
• 🔄 Template patching to pull upstream improvements into existing projects

Installation

Recommended: Install with pipx for isolated command-line tools:

pipx install fips-agents-cli

pipx installs the CLI in an isolated environment while making it globally available. This prevents dependency conflicts with other Python packages.

Alternative: Using pip

If you prefer pip or don't have pipx installed:

pip install fips-agents-cli

Note: Consider using a virtual environment to avoid dependency conflicts.

Alternative: Install from source (for development)

For contributing or local development:

# Clone the repository
git clone https://github.com/fips-agents/fips-agents-cli.git
cd fips-agents-cli

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install in editable mode with dev dependencies
pip install -e .[dev]

Quick Start

Create projects

# MCP server
fips-agents create mcp-server my-mcp-server

# AI agent
fips-agents create agent my-research-agent

# Go HTTP gateway (proxies to an agent backend)
fips-agents create gateway my-gateway

# Chat UI (connects to a gateway or agent)
fips-agents create ui my-chat-ui

# Code execution sandbox (sidecar for agents)
fips-agents create sandbox my-sandbox

# ModelCar (HuggingFace model as container)
fips-agents create model-car ibm-granite/granite-3.1-2b-instruct \
    quay.io/user/models:granite-3.1-2b-instruct

Generate components in an existing MCP server project

cd my-mcp-server

fips-agents generate tool search_documents --description "Search through documents"
fips-agents generate resource config_data --description "Application configuration"
fips-agents generate prompt code_review --description "Review code for best practices"
fips-agents generate middleware auth_middleware --description "Authentication middleware"

Check for template updates

cd my-mcp-server

fips-agents patch check
fips-agents patch all --dry-run

Command Reference

Help

fips-agents --version
fips-agents --help
fips-agents add --help
fips-agents create --help
fips-agents generate --help
fips-agents patch --help

---

Create Commands

The create command group scaffolds new projects from templates.

Shared Options (mcp-server, agent, gateway, ui)

All create subcommands (except model-car) accept the same options:

Option	Description
--target-dir, -t PATH	Target directory for the project (default: current directory)
--no-git	Skip git repository initialization
--github	Create GitHub repository and push code
--local	Create local project only (skip GitHub)
--yes, -y	Non-interactive mode (use defaults, skip prompts)
--private	Make GitHub repository private (default: public)
--org TEXT	GitHub organization to create repository in
--description, -d TEXT	GitHub repository description
--remote-only	Create GitHub repo only, don't clone locally

When gh CLI is installed and authenticated, the tool prompts interactively about GitHub repo creation. Use --local to suppress this, --github to require it, or --yes for fully non-interactive operation.

create mcp-server

fips-agents create mcp-server <project-name> [OPTIONS]

Creates an MCP server project from the mcp-server-template repository. The template includes FastMCP server bootstrap, component auto-discovery, Jinja2 generator templates, Makefile, Containerfile, and tests.

Arguments:

• project-name — Name for your MCP server project

Examples:

# Create in current directory
fips-agents create mcp-server my-mcp-server

# Create in specific directory without git
fips-agents create mcp-server my-server -t ~/projects --no-git

# Create with private GitHub repo in an organization
fips-agents create mcp-server my-server --github --private --org my-org

# Non-interactive, local only
fips-agents create mcp-server my-server --yes --local

# Create GitHub repo without cloning locally
fips-agents create mcp-server my-server --remote-only --org my-org

create agent

fips-agents create agent <project-name> [OPTIONS]

Creates an AI agent project from the agent-template monorepo (templates/agent-loop/ subdirectory). The template includes an agent loop skeleton, Helm chart, Makefile, Containerfile, and AGENTS.md.

Arguments:

• project-name — Name for your agent project

Options: Same as create mcp-server (see shared options table above).

Examples:

# Create agent project
fips-agents create agent my-research-agent

# Create with GitHub repo in an organization
fips-agents create agent my-agent --github --org fips-agents

# Non-interactive mode
fips-agents create agent my-agent --yes --local

create gateway

fips-agents create gateway <project-name> [OPTIONS]

Creates a Go HTTP gateway project from the gateway-template repository. The gateway proxies OpenAI-compatible /v1/chat/completions requests to an agent backend, with SSE streaming support, heartbeat keepalives, health/readiness probes, and an A2A agent discovery card.

Arguments:

• project-name — Name for your gateway project

Options: Same shared options as above.

Examples:

# Create gateway project
fips-agents create gateway my-gateway

# Create with GitHub repo
fips-agents create gateway my-gateway --github --org my-org

create ui

fips-agents create ui <project-name> [OPTIONS]

Creates a chat UI project from the ui-template repository. A Go server with embedded HTML/CSS/JS that provides a browser-based chat interface. Includes a built-in reverse proxy to the backend API, SSE streaming, and markdown rendering.

Arguments:

• project-name — Name for your UI project

Options: Same shared options as above.

Examples:

# Create UI project
fips-agents create ui my-chat-ui

# Create with GitHub repo
fips-agents create ui my-chat-ui --github --private

create sandbox

fips-agents create sandbox <project-name> [OPTIONS]

Creates a code execution sandbox project from the code-sandbox repository. The sandbox provides a FastAPI-based sidecar for secure code execution inside agent pods, with multiple language profiles (base, data-science).

Arguments:

• project-name -- Name for your sandbox project

Options: Same shared options as above.

Examples:

# Create sandbox project
fips-agents create sandbox my-sandbox

# Create with GitHub repo
fips-agents create sandbox my-sandbox --github --private

# Non-interactive mode
fips-agents create sandbox my-sandbox --yes --local

create model-car

fips-agents create model-car <hf-repo> <quay-uri> [OPTIONS]

Creates a ModelCar project for packaging a HuggingFace model as a container image. Generates download scripts, Containerfile, build-and-push scripts, and project metadata.

Arguments:

• hf-repo — HuggingFace repo URL or ID (e.g., ibm-granite/granite-3.1-2b-instruct or the full https://huggingface.co/... URL)
• quay-uri — Container registry URI with tag (e.g., quay.io/user/models:granite-3.1-2b-instruct)

Options:

Option	Description
--target-dir, -t PATH	Target directory for the project (default: current directory)

Examples:

# Using repo ID
fips-agents create model-car ibm-granite/granite-3.1-2b-instruct \
    quay.io/wjackson/models:granite-3.1-2b-instruct

# Using full URL with custom target directory
fips-agents create model-car https://huggingface.co/ibm-granite/granite-3.1-2b-instruct \
    quay.io/wjackson/models:granite-3.1-2b-instruct -t ~/models

---

Generate Commands

The generate command group creates individual components (tools, resources, prompts, middleware) in existing projects.

Run these commands from within your MCP server project directory. The CLI locates the project root by looking for pyproject.toml with a fastmcp dependency.

Each generator:

1. Validates the component name and checks for conflicts
2. Loads Jinja2 templates from .fips-agents-cli/generators/ in your project
3. Renders both the component file and a corresponding test file
4. Validates Python syntax on the generated code
5. Runs the generated tests automatically

Component names must be valid Python identifiers in snake_case. Subdirectory paths (e.g., country-profiles/japan) are supported for resources.

generate tool

fips-agents generate tool <name> [OPTIONS]

Generates a tool component with FastMCP @mcp.tool() decorator.

Options:

Option	Description
--async/--sync	Generate async or sync function (default: async)
--with-context	Include FastMCP Context parameter
--with-auth	Include authentication decorator
--scopes TEXT	Required OAuth scopes, comma-separated (e.g., "read:data,write:data")
--with-elicitation	Include elicitation example code
--description, -d TEXT	Tool description
--params PATH	JSON file with parameter definitions
--read-only	Mark as read-only operation (default: true)
--idempotent	Mark as idempotent (default: true)
--open-world	Mark as open-world operation
--return-type TEXT	Return type annotation (default: str)
--dry-run	Show what would be generated without creating files

Examples:

# Basic tool
fips-agents generate tool search_documents --description "Search through documents"

# Tool with context and authentication
fips-agents generate tool fetch_user_data --with-context --with-auth --scopes "read:users"

# Tool with parameters from JSON file
fips-agents generate tool advanced_search --params params.json

# Sync tool with custom return type
fips-agents generate tool process_data --sync --return-type "dict[str, Any]"

# Preview without creating files
fips-agents generate tool test_tool --description "Test" --dry-run

generate resource

fips-agents generate resource <name> [OPTIONS]

Generates a resource component with FastMCP @mcp.resource() decorator. Supports subdirectory paths (e.g., country-profiles/japan creates src/resources/country_profiles/japan.py). URI template parameters (e.g., {id}) are automatically extracted as function arguments.

Options:

Option	Description
--async/--sync	Generate async or sync function (default: async)
--with-context	Include FastMCP Context parameter
--description, -d TEXT	Resource description
--uri TEXT	Resource URI (default: resource://<name>)
--mime-type TEXT	MIME type (default: text/plain)
--dry-run	Show what would be generated without creating files

Examples:

# Basic resource
fips-agents generate resource config_data --description "Application configuration"

# Resource with URI template (parameters auto-extracted as function args)
fips-agents generate resource user_profile --uri "resource://users/{id}"

# Resource in a subdirectory
fips-agents generate resource country-profiles/japan --description "Japan country profile"

# Resource with specific MIME type
fips-agents generate resource json_config --mime-type "application/json"

generate prompt

fips-agents generate prompt <name> [OPTIONS]

Generates a prompt component with FastMCP @mcp.prompt() decorator. Note: prompts default to sync (unlike tools and resources which default to async).

Options:

Option	Description
--async/--sync	Generate async or sync function (default: sync)
--with-context	Include FastMCP Context parameter
--description, -d TEXT	Prompt description
--params PATH	JSON file with parameter definitions
--return-type [str|Message|list[Message]]	Return type (default: str)
--with-schema	Include JSON schema example in prompt body
--prompt-name TEXT	Override decorator name (default: function name)
--title TEXT	Human-readable title for the prompt
--tags TEXT	Comma-separated tags for categorization
--disabled	Generate prompt in disabled state
--meta TEXT	JSON string of metadata (e.g., '{"version": "1.0"}')
--dry-run	Show what would be generated without creating files

Examples:

# Basic prompt
fips-agents generate prompt code_review --description "Review code for best practices"

# Prompt with parameters and schema
fips-agents generate prompt summarize_text --params params.json --with-schema

# Async prompt returning a Message
fips-agents generate prompt fetch_data --async --with-context --return-type Message

# Prompt with metadata and tags
fips-agents generate prompt generate_report \
    --prompt-name "report_generator" \
    --title "Report Generator" \
    --tags "reporting,analysis" \
    --meta '{"version": "2.0", "author": "data-team"}'

generate middleware

fips-agents generate middleware <name> [OPTIONS]

Generates a middleware component using FastMCP's @mcp.middleware() decorator (v3.x class-based pattern). Middleware wraps tool execution and receives a Context and next_handler.

Options:

Option	Description
--async/--sync	Generate async or sync function (default: async)
--hook-type [before_tool|after_tool|on_error]	Scaffold a specific hook pattern (default: general wrapper)
--description, -d TEXT	Middleware description
--dry-run	Show what would be generated without creating files

Examples:

# General-purpose wrapper
fips-agents generate middleware auth_middleware --description "Authentication middleware"

# Before-tool hook
fips-agents generate middleware rate_limiter --hook-type before_tool

# Error handler (sync)
fips-agents generate middleware error_reporter --hook-type on_error --sync

Parameters JSON Schema

When using --params with generate tool or generate prompt, provide a JSON file defining parameter names, types, and constraints:

[
  {
    "name": "query",
    "type": "str",
    "description": "Search query",
    "required": true,
    "min_length": 1,
    "max_length": 100
  },
  {
    "name": "limit",
    "type": "int",
    "description": "Maximum results to return",
    "required": false,
    "default": 10,
    "ge": 1,
    "le": 100
  }
]

Supported Types:

• str, int, float, bool
• list[str], list[int], list[float]
• Optional[str], Optional[int], Optional[float], Optional[bool]

Pydantic Field Constraints:

• min_length, max_length (for strings)
• ge, le, gt, lt (for numbers)
• pattern (for regex validation on strings)
• default (default value when optional)

---

Patch Commands

The patch command group updates files in existing MCP server projects from the upstream template repository without overwriting your custom code. It shows interactive diffs for files that may contain customizations.

Run these commands from within your project directory.

patch check

fips-agents patch check

Check for available template updates. Shows what files have changed in the template since your project was created, organized by category.

patch generators

fips-agents patch generators [--dry-run]

Update code generator templates (Jinja2 templates in .fips-agents-cli/generators/).

patch core

fips-agents patch core [--dry-run]

Update core infrastructure files (loaders, server bootstrap).

patch docs

fips-agents patch docs [--dry-run]

Update documentation files and examples.

patch build

fips-agents patch build [--dry-run]

Update build and deployment files (Makefile, Containerfile).

patch all

fips-agents patch all [--dry-run] [--skip-confirmation]

Update all patchable file categories at once. Prompts for confirmation before starting unless --skip-confirmation is passed.

All patch subcommands (except check) accept --dry-run to preview changes without modifying files.

---

Add Commands

The add command group wires optional capabilities into existing agent projects created with fips-agents create agent.

Run these commands from within your agent project directory. The CLI locates the project root by looking for agent.yaml.

add code-executor

fips-agents add code-executor

Adds the code_executor tool to your agent's tools/ directory and enables the sandbox sidecar in chart/values.yaml. The tool sends Python code to a sandbox container for isolated execution.

What it does:

1. Writes tools/code_executor.py with the sandbox client tool
2. Sets sandbox.enabled: true in chart/values.yaml (if the section exists)
3. Prints deployment guidance for sidecar and remote service modes

Deployment modes:

• Sidecar (default): The sandbox runs as a container in the same pod. The tool connects to http://localhost:8000.
• Remote service: The sandbox runs as a separate deployment. Set SANDBOX_URL to point to it.

Example:

cd my-research-agent
fips-agents add code-executor

Project Name Requirements

Project names must follow these rules:

• Start with a lowercase letter
• Contain only lowercase letters, numbers, hyphens (-), and underscores (_)
• Not be empty

Valid examples: my-server, test_mcp, server123, my-awesome-mcp-server

Invalid examples: MyServer (uppercase), 123server (starts with number), my@server (special characters)

After Creating a Project

MCP Server

cd my-mcp-server
python -m venv venv
source venv/bin/activate
pip install -e .[dev]
pytest
# Start developing in src/my_mcp_server/

AI Agent

cd my-research-agent
python -m venv venv
source venv/bin/activate
pip install -e .[dev]
pytest
# See AGENTS.md for the /plan-agent slash command workflow

Gateway

cd my-gateway
make build           # Build the Go binary
make run             # Run locally (set BACKEND_URL to your agent)
make build-openshift PROJECT=my-gateway  # Build on OpenShift
make deploy PROJECT=my-gateway           # Deploy via Helm

Chat UI

cd my-chat-ui
make build           # Build the Go binary
make run             # Run locally (set API_URL to your gateway/agent)
make build-openshift PROJECT=my-chat-ui  # Build on OpenShift
make deploy PROJECT=my-chat-ui           # Deploy via Helm

Sandbox

cd my-sandbox
make install             # Install dependencies
make test                # Run tests
make build               # Build container
make build PROFILE=data-science  # Build with profile

ModelCar

cd granite-3.1-2b-instruct
./download.sh        # Download model weights
./build-and-push.sh  # Build container and push to registry

Development

Setup Development Environment

# Clone the repository
git clone https://github.com/fips-agents/fips-agents-cli.git
cd fips-agents-cli

# Create virtual environment
python -m venv venv
source venv/bin/activate

# Install with dev dependencies
pip install -e .[dev]

Running Tests

# Run all tests
pytest

# Run with coverage
pytest --cov=fips_agents_cli --cov-report=html

# Run specific test file
pytest tests/test_create.py

# Run specific test
pytest tests/test_create.py::TestCreateMcpServer::test_successful_creation

Code Quality

# Format code with Black
black src tests

# Lint with Ruff
ruff check src tests

# Type checking (if using mypy)
mypy src

Project Structure

fips-agents-cli/
├── pyproject.toml          # Project configuration
├── README.md               # This file
├── CLAUDE.md               # Claude Code developer instructions
├── RELEASE_CHECKLIST.md    # Release procedures
├── docs/                   # User-facing documentation
├── planning/               # Design and implementation plans
├── research/               # Technology evaluations
├── retrospectives/         # Post-effort retrospectives
├── scripts/                # Automation scripts
├── src/
│   └── fips_agents_cli/
│       ├── __init__.py
│       ├── __main__.py     # Entry point for python -m
│       ├── cli.py          # Main CLI application
│       ├── version.py      # Version information
│       ├── commands/       # CLI command implementations
│       │   ├── add.py      # add code-executor (wire capabilities)
│       │   ├── create.py   # create mcp-server, agent, gateway, ui
│       │   ├── generate.py # generate tool/resource/prompt/middleware
│       │   ├── model_car.py # create model-car
│       │   └── patch.py    # patch command
│       └── tools/          # Utility modules
│           ├── filesystem.py
│           ├── git.py
│           ├── github.py
│           ├── project.py
│           └── validation.py
└── tests/
    ├── conftest.py         # Pytest fixtures
    ├── test_create.py      # Create command tests
    ├── test_project.py     # Project utilities tests
    └── ...

Documentation

• docs/ — User-facing reference: publishing guides, release procedures
• planning/ — Design documents and implementation plans for upcoming features
• research/ — Technology evaluations and architectural analysis
• retrospectives/ — Post-effort retrospectives

See docs/README.md for a full index.

Requirements

• Python 3.10 or higher
• Git (for cloning templates and initializing repositories)

Dependencies

• click (>=8.1.0): Command-line interface creation
• rich (>=13.0.0): Terminal output formatting
• gitpython (>=3.1.0): Git operations
• tomlkit (>=0.12.0): TOML file manipulation
• jinja2 (>=3.1.2): Template rendering for component generation

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Make your changes
4. Run tests (pytest)
5. Format code (black src tests)
6. Commit your changes (git commit -m 'Add amazing feature')
7. Push to the branch (git push origin feature/amazing-feature)
8. Open a Pull Request

Troubleshooting

Git not found

If you see "Git is not installed" error:

• macOS: brew install git
• Linux: sudo apt-get install git or sudo yum install git
• Windows: Download from https://git-scm.com/downloads

Directory already exists

If you see "Directory already exists" error:

• Choose a different project name
• Remove the existing directory: rm -rf project-name
• Use a different target directory with --target-dir

Template clone fails

If template cloning fails:

• Check your internet connection
• Verify the template repository is accessible: https://github.com/fips-agents/mcp-server-template
• Try again later if GitHub is experiencing issues

Agent template clone fails

If cloning the agent template fails:

• The agent template is in a monorepo at https://github.com/fips-agents/agent-template
• Verify the repository is accessible and the templates/agent-loop/ subdirectory exists
• Check that you have access to the organization's repositories

ModelCar validation errors

• Invalid HuggingFace repo: Use either a full URL (https://huggingface.co/org/model) or a short ID (org/model)
• Invalid Quay URI: Must include a tag (e.g., quay.io/user/repo:tag), not just quay.io/user/repo
• Registry login: The CLI checks for podman login status; run podman login quay.io if prompted

"Not in an MCP server project directory"

When using generate or patch commands:

• Ensure you're running the command from within an MCP server project
• Check that pyproject.toml exists with fastmcp dependency
• If the project wasn't created with fips-agents create mcp-server, generator templates may be missing

"Component already exists"

If you see this error:

• Choose a different component name
• Manually remove the existing component file from src/<component-type>/
• Check the component type directory for existing files

Invalid component name

Component names must:

• Be valid Python identifiers (snake_case)
• Not be Python keywords (for, class, etc.)
• Start with a letter or underscore
• Contain only letters, numbers, and underscores

License

MIT License - see LICENSE file for details

Links

• Repository: https://github.com/fips-agents/fips-agents-cli
• Issues: https://github.com/fips-agents/fips-agents-cli/issues
• MCP Protocol: https://modelcontextprotocol.io/

Changelog

Version 0.7.0

• Feature: New add command group for composable agent capabilities
• Feature: fips-agents add code-executor wires sandbox code execution into existing agent projects (tool + Helm chart sidecar config)
• Improvement: Updated generate group description to reflect broader scope beyond MCP

Version 0.6.2

• Fix: Go project scaffolding now updates import paths in all .go source files
• Fix: Go project scaffolding now updates Helm template references (chart/templates/*.yaml)

Version 0.6.1

• Chore: Updated all project URLs from rdwj/ to fips-agents/ org
• Chore: Post-transfer PyPI publishing verification release

Version 0.6.0

• Feature: Added create sandbox command for scaffolding code execution sandbox projects from fips-agents/code-sandbox template
• Feature: All template URLs consolidated under github.com/fips-agents/ organization

Version 0.5.1

• Fix: SSE streaming tool call ID tracking by call_id not array index

Version 0.5.0

• Feature: Added create gateway command for scaffolding Go HTTP gateway projects from the gateway-template
• Feature: Added create ui command for scaffolding chat UI projects from the ui-template
• Feature: Go project customization pipeline (customize_go_project) handles go.mod, Helm charts, Containerfile, Makefile, and static HTML
• Fix: Agent scaffolding now replaces agent-template in all Helm chart templates, preventing resource name collisions
• Fix: Agent scaffolding now replaces OWNER/REPO placeholder in Containerfile with actual GitHub repo

Version 0.4.0

• Feature: Added create agent command for scaffolding AI agent projects from the agent-loop template
• Feature: New clone_template_subdir() utility for extracting templates from monorepos
• Feature: Agent project customization across pyproject.toml, Helm chart, Makefile, Containerfile, and AGENTS.md
• Improvement: Reorganized documentation into docs/, planning/, and research/ directories

Version 0.3.0

• Feature: Added --hook-type option to middleware generator with before_tool, after_tool, and on_error choices for v3.x template scaffolding
• Fix: Fixed brittle test mocking for GitHub CLI detection in create command tests

Version 0.2.1

• Fix: Update CLI for FastMCP 3.x template compatibility (auth.py replaces loaders.py, updated prompt return types)
• Fix: Reorder create mcp-server precondition checks to run before interactive GitHub prompt, so errors like missing git or existing directories fail fast without prompting first

Version 0.2.0

• Feature: GitHub integration for create mcp-server command
• Feature: New --github flag to create GitHub repository and push code
• Feature: New --local flag to skip GitHub and create local-only project
• Feature: Non-interactive mode (--yes) for agent/CI workflows
• Feature: New --private flag to create private GitHub repositories
• Feature: New --org option to create repositories in GitHub organizations
• Feature: New --description option for GitHub repository descriptions
• Feature: New --remote-only flag to create GitHub repo without local clone
• Feature: Added missing patch build subcommand for updating build/deployment files
• Feature: GitHub metadata tracking in .template-info file
• Feature: New git utilities for remote management (add_remote, push_to_remote)
• Improvement: Auto-detects gh CLI and prompts user when available
• Improvement: Clean git history - customizes project before initial push

Version 0.1.9

• Feature: Added .fips-agents-cli directory to ModelCar projects with generation metadata
• Feature: ModelCar projects now include info.json with source, destination, and generator info
• Feature: ModelCar projects now include CLAUDE.md with project-specific Claude Code instructions
• Improvement: Consistent project metadata structure between MCP server and ModelCar projects

Version 0.1.8

• Feature: Added GitHub Copilot agent template support
• Feature: Added model-car command for creating model car projects
• Fix: Subdirectory names now convert hyphens to underscores for valid Python package names
• Fix: Added module_path template variable for proper import paths in subdirectory components
• Fix: Script generator improvements
• Improvement: Removed download filter for cleaner template handling
• Improvement: Added info file tracking for generated projects

Version 0.1.7

• Critical Fix: Resource generator now automatically extracts URI template parameters (e.g., {country_code}, {id}) and adds them to function signature
• Critical Fix: Generated resources now comply with FastMCP requirement that URI parameters must match function arguments
• Fix: Updated test templates to use .fn attribute pattern for testing FastMCP decorated functions
• Fix: Improved component_exists() to properly handle subdirectory paths
• Fix: Removed obsolete _preview_prompt_utility.py from template (YAML-based prompts removed)
• Improvement: Generated resource tests now pass out of the box
• Improvement: Resource function signatures now include proper type hints and docstring Args sections

Version 0.1.6

• Feature: Added fips-agents patch command for selective template updates without losing custom code
• Feature: Resource subdirectory support - organize resources in hierarchies (e.g., country-profiles/japan)
• Feature: generate resource now supports subdirectory paths for better organization
• Improvement: Recursive resource loading from subdirectories with automatic discovery
• Improvement: Patch command shows interactive diffs for files that may be customized
• Improvement: Template version tracking enables smart updates via .template-info file
• Enhancement: Cleaned up redundant auto-discovery code from template __init__.py files

Version 0.1.5

• Feature: Added .template-info file to track CLI version and template commit hash in generated projects
• Improvement: Implemented auto-discovery for tools/resources/prompts/middleware components, eliminating manual registration in __init__.py files
• Improvement: Template components can now be added or removed by simply creating/deleting files—no registry updates required

Version 0.1.4

• Fix: Fixed Containerfile in template to remove incorrect prompts/ copy line
• Fix: Fixed CLI to correctly update entry point script names for new projects
• Fix: Updated CLI to handle new multi-module template structure (core/, tools/, etc.)
• Fix: Updated fallback project name to match current template (fastmcp-unified-template)
• Improvement: Improved messaging when template uses multi-module structure

Version 0.1.3

• Improvement: Enhanced prompt creation to comply with FastMCP 2.9+ requirements

Version 0.1.2

• Documentation: Updated documentation and improved release management guidance
• Automation: Added /create-release slash command for automated release workflow
• Automation: Created scripts/release.sh for streamlined version management and tagging
• Documentation: Added RELEASE_CHECKLIST.md with comprehensive release procedures

Version 0.1.1

• Added fips-agents generate command group
• Component generation: tools, resources, prompts, middleware
• Jinja2-based template rendering
• Parameter validation and JSON schema support
• Auto-run pytest on generated components
• Dry-run mode for previewing changes
• Comprehensive error handling and validation

Version 0.1.0 (MVP)

• Initial release
• fips-agents create mcp-server command
• Template cloning and customization
• Git repository initialization
• Comprehensive test suite
• Beautiful CLI output with Rich