IAM Policy Autopilot
An open source Model Context Protocol (MCP) server and command-line tool that helps your AI coding assistants quickly create baseline IAM policies that you can refine as your application evolves, so you can build faster. IAM Policy Autopilot analyzes your application code locally to generate identity-based policies for application roles, enabling faster IAM policy creation and reducing access troubleshooting time. IAM Policy Autopilot supports applications built in Python, Go, and TypeScript.
Table of Contents
- Who is IAM Policy Autopilot for?
- How is IAM Policy Autopilot helpful?
- Best Practices and Considerations
- Getting Started
- CLI Usage
- Build Instructions
- Workspace Structure
- Development
- Security
- License
Who is IAM Policy Autopilot for?
IAM Policy Autopilot is for builders on AWS using AI coding assistants, including developers, product managers, technical experimenters, and business leaders.
How is IAM Policy Autopilot helpful?
IAM Policy Autopilot is:
Fast
IAM Policy Autopilot accelerates development by generating baseline identity-based IAM policies. Your AI coding assistant can call IAM Policy Autopilot to analyze AWS SDK calls within your application. IAM Policy Autopilot then automatically creates the baseline IAM permissions for your application roles.
Reliable
IAM Policy Autopilot's deterministic code analysis helps create reliable and valid IAM policies that reduce policy troubleshooting. By using valid policies created with the MCP server, you reduce time spent on policy-related debugging and accelerate application deployment by avoiding permission-related delays.
Up-to-date
IAM Policy Autopilot stays up to date with the latest AWS services and features so that builders and coding assistants have access to the latest AWS IAM permissions knowledge. It helps keep your application role's permissions current with AWS's evolving capabilities.
Best Practices and Considerations
Review and refine policies generated by IAM Policy Autopilot
IAM Policy Autopilot generates baseline policies to provide a starting point that you can refine as your application matures. Review the generated policies to ensure they align with your security requirements before deploying them. Use the --explain feature with action patterns (e.g., --explain 's3:*') to understand which operations led to an action being included in the generated policies.
Understand the IAM Policy Autopilot scope
IAM Policy Autopilot produces IAM identity-based policies, but doesn't support resource-based policies such as S3 bucket policies or KMS key policies, Resource Control Policies (RCPs), Service Control Policies (SCPs), and permission boundaries. These are the limitations that you need to keep in mind. For example, if your code calls s3.getObject(bucketName) where bucketName is determined at runtime, IAM Policy Autopilot currently doesn't predict which bucket will be accessed.
Understand the boundary between IAM Policy Autopilot and your coding assistant
IAM Policy Autopilot generates policies with specific actions based on deterministic analysis of your code. When you use the MCP server integration, your AI coding assistant receives this policy and might modify it when creating infrastructure-as-code templates. For example, you might see the assistant add specific resource Amazon Resource Names (ARNs) or include KMS key IDs based on additional context from your code. These changes come from your coding assistant's interpretation of your broader code context, not from the static analysis provided by IAM Policy Autopilot. Always review content generated by your coding assistant before deployment to verify that it meets your security requirements.
Use service hints for accurate policies
IAM Policy Autopilot's static analysis may include permissions for AWS services your application doesn't use. This happens when method names in your code match AWS SDK calls from multiple services. For example, a method called listAccounts() might generate permissions for both AWS Organizations and Amazon Chime services.
Recommended approach: Use the --service-hints option to specify only the AWS services your application actually uses. This helps IAM Policy Autopilot scope down which SDK calls to analyze, but the final policy may still include actions from other services if they're required by the operations you perform:
# More accurate - specify only services you use
iam-policy-autopilot generate-policies ./src/app.py --service-hints s3 iam organizations --pretty
# Less accurate - may include unnecessary permissions
iam-policy-autopilot generate-policies ./src/app.py --pretty
This significantly reduces unnecessary permissions and generates more targeted policies. Note that the final policy may still include actions from services not in your hints if they're required for the operations you perform (e.g., KMS actions for S3 encryption).
Note: When using the MCP server integration with AI coding assistants, the assistant is expected to automatically provide appropriate service hints based on your code context. The --service-hints option is primarily for CLI usage.
Getting Started
Installation
Option 1: Using uv (Recommended)
Install uv from Astral.
No additional installation needed - you can run IAM Policy Autopilot directly using uvx iam-policy-autopilot.
Option 2: Using pip
Install pip.
pip install iam-policy-autopilot
Option 3: Direct installation (MacOS/Linux only)
To install the latest release directly, run the following script to download and install as a system utility.
curl -sSL https://github.com/awslabs/iam-policy-autopilot/raw/refs/heads/main/install.sh | sudo sh
This will install the latest release directly to /usr/local/bin/iam-policy-autopilot.
AWS Configuration
IAM Policy Autopilot requires AWS credentials to apply policy fixes and upload policies for AccessDenied debugging.
Install AWS CLI and configure your AWS credentials.
For more information on AWS credential configuration, see the AWS CLI Configuration Guide.
MCP Server Configuration
Configure the MCP server in your MCP client configuration to enable your AI coding assistant to generate IAM policies.
For Kiro
Get Kiro from https://kiro.dev/
If using uv/uvx:
Add the following configuration to your project-level .kiro/settings/mcp.json:
{
"mcpServers": {
"iam-policy-autopilot": {
"command": "uvx",
"args": ["iam-policy-autopilot", "mcp-server"],
"env": {
"AWS_PROFILE": "your-profile-name",
"AWS_REGION": "us-east-1"
},
"disabled": false,
"autoApprove": []
}
}
}
If using pip:
{
"mcpServers": {
"iam-policy-autopilot": {
"command": "iam-policy-autopilot",
"args": ["mcp-server"],
"env": {
"AWS_PROFILE": "your-profile-name",
"AWS_REGION": "us-east-1"
},
"disabled": false,
"autoApprove": []
}
}
}
For Kiro CLI
Get Kiro CLI from https://kiro.dev/cli
Kiro Cli uses the same configuration as Kiro mentioned above, additionally, MCPs for Kiro CLI can also be setup via:
If using uv/uvx:
kiro-cli mcp add \
--name iam-policy-autopilot \
--command "uvx" \
--args "iam-policy-autopilot","mcp-server"
If using pip:
kiro-cli mcp add \
--name iam-policy-autopilot \
--command "iam-policy-autopilot" \
--args "mcp-server"
For Claude Desktop
Add to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json
If using uv/uvx:
{
"mcpServers": {
"iam-policy-autopilot": {
"command": "uvx",
"args": ["iam-policy-autopilot", "mcp-server"],
"env": {
"AWS_PROFILE": "your-profile-name",
"AWS_REGION": "us-east-1"
}
}
}
}
If using pip:
{
"mcpServers": {
"iam-policy-autopilot": {
"command": "iam-policy-autopilot",
"args": ["mcp-server"],
"env": {
"AWS_PROFILE": "your-profile-name",
"AWS_REGION": "us-east-1"
}
}
}
}
Kiro Power Configuration
IAM Policy Autopilot has an associated Kiro power configuration inside the power-iam-policy-autopilot directory. This can be used to install a corresponding Kiro power in your Kiro editor.
Enabling the IAM Policy Autopilot Kiro Power
To enable the IAM Policy Autopilot Kiro Power, first install the uv package manager by following these instructions. Then, do the following steps within Kiro:
- Go to the "Powers" menu in the menubar on the left-hand-side.
- Click
Add Custom Power->Import power from Github - In the text prompt that then appears, enter
https://github.com/awslabs/iam-policy-autopilot/tree/main/power-iam-policy-autopilot. - Kiro should automatically install a new Kiro power called
IAM Policy Autopilotwithin your Kiro code editor. This power should be visible in thePowersmenu.
If the above steps for installing the power from a GitHub repository URL does not work, you can also clone the repository and import the power directly, by doing the following:
- Clone the git repository
https://github.com/awslabs/iam-policy-autopilot, and remember the directory to where you cloned the repo. - Go to the "Powers" menu in the menubar on the left-hand-side.
- Click
Add Custom Power->Import power from a folder - In the text prompt that then appears, select the
power-iam-policy-autopilotfolder in your cloned repository. For instance, if the repository is cloned to~/workplace/iam-policy-autopilot, you should select or enter~/workplace/iam-policy-autopilot/power-iam-policy-autopilot. - Kiro should automatically install a new Kiro power called
IAM Policy Autopilotwithin your Kiro code editor. This power should be visible in thePowersmenu.
Why use IAM Policy Autopilot's Kiro Power?
Kiro powers generally offer a more refined experience than traditional MCP servers because they enable MCP tools to be loaded more selectively & deliberately, reducing LLM token usage and avoiding LLM context overcrowding.
IAM Policy Autopilot's Kiro power specifically enhances the traditional MCP experience, for multiple reasons:
- This Kiro Power provides your LLM agent with more steering guidance, offering it more information on the specific use cases and best practices of our MCP tooling.
- This Kiro power prompts your LLM agent to give a tutorial of the MCP tools offered by IAM Policy Autopilot, allowing you to better understand how our MCP tooling assists your use case.
- This Kiro Power provides your LLM agent with step-by-step onboarding validation, allowing it to detect any problems with installations and provide remediation steps for those problems.
CLI Usage
The iam-policy-autopilot CLI tool provides three main commands:
Generate IAM policies from source code and fix AccessDenied errors
Usage: iam-policy-autopilot <COMMAND>
Commands:
fix-access-denied Fix AccessDenied errors by analyzing and optionally applying IAM policy changes
generate-policies Generates complete IAM policy documents from source files
mcp-server Start MCP server
help Print this message or the help of the given subcommand(s)
Options:
-h, --help Print help (see more with '--help')
-V, --version Print version
Commands
generate-policies - Generates complete IAM policy documents from source files
iam-policy-autopilot generate-policies <source_files> [OPTIONS]
Example:
iam-policy-autopilot generate-policies \
./src/app.py \
--region us-east-1 \
--account 123456789012 \
--pretty
Options:
--region <REGION>- AWS region for resource ARNs--account <ACCOUNT>- AWS account ID for resource ARNs--service-hints <SERVICES>- Limit analysis to only the services your application actually uses if you know them. This helps reduce unnecessary permissions.--upload-policies <PREFIX>- Upload generated policies to AWS IAM with the specified prefix--pretty- Pretty-print JSON output
fix-access-denied - Fix AccessDenied errors by analyzing and optionally applying IAM policy changes
iam-policy-autopilot fix-access-denied <access-denied-error-message> [OPTIONS]
Example:
iam-policy-autopilot fix-access-denied \
"User: arn:aws:iam::123456789012:user/test is not authorized to perform: s3:GetObject on resource: arn:aws:s3:::my-bucket/file.txt"
Options:
--yes- Auto-apply policy changes without confirmation
mcp-server - Start MCP server locally
iam-policy-autopilot mcp-server [OPTIONS]
Options:
--transport <TRANSPORT>- Transport type:stdio(default) orhttp
Example with HTTP transport:
# Start server at http://127.0.0.1:8001/mcp
iam-policy-autopilot mcp-server --transport http
Build Instructions
Prerequisites
Setup
Clone the repository with submodules:
git clone --recurse-submodules https://github.com/awslabs/iam-policy-autopilot.git
cd iam-policy-autopilot
Build the project:
cargo build --release
The compiled binary will be located at target/release/iam-policy-autopilot.
Using the Built Binary with MCP
If you build from source, you can configure MCP clients to use the compiled binary:
{
"mcpServers": {
"iam-policy-autopilot": {
"command": "/path/to/iam-policy-autopilot",
"args": ["mcp-server"]
}
}
}
Workspace Structure
This workspace contains several crates that work together:
iam-policy-autopilot-policy-generation/- Core library providing SDK extraction and enrichment capabilitiesiam-policy-autopilot-access-denied/- Core library for parsing AccessDenied errors and synthesizing IAM policiesiam-policy-autopilot-tools/- Policy upload utilities and AWS integration toolsiam-policy-autopilot-cli/- Unified CLI tool providing all commandsiam-policy-autopilot-mcp-server/- MCP server integration for IDE and tool integration
Development
Running Tests
# Run all tests
cargo test --workspace
# Run tests for specific crate
cargo test -p iam-policy-autopilot-cli
cargo test -p iam-policy-autopilot-access-denied
cargo test -p iam-policy-autopilot-policy-generation
# Run integration tests
cargo test -p iam-policy-autopilot-cli --test integration_tests
Building Release Version
cargo build --release
The compiled binary will be located at target/release/iam-policy-autopilot.
Security
See CONTRIBUTING for more information.
License
This project is licensed under the Apache-2.0 License.