aws-iam Skill (AWS) for Claude: What It Does and How to Install It

Makes Claude avoid common IAM mistakes and draft baseline IAM policies and roles from your code or Terraform plan.

Installs: 10.5K; on skills.sh
Stars: 3.4K; on GitHub
Files: 4; besides SKILL.md
Updated: Oct 2026; last repo commit

This is Amazon Web Services' skill for IAM. SKILL.md lists verified corrections for IAM details that AI agents often get wrong, in 10 groups from CloudTrail and STS to SAML, MFA and policy evaluation. It also points to guides for role management and for generating policies.

When you give Claude application source code or a Terraform plan JSON, the policy guide tells it to propose the uvx iam-policy-autopilot@latest generate-policies command instead of writing the policy by hand. The README says you need uv installed, and AWS credentials configured locally for API calls.

Good for

Developers writing IAM roles and policies for AWS apps
Teams generating least-privilege policies from source code
Platform engineers working with STS, Organizations and SAML
Anyone checking conditions, trust policies and bucket policies

Install aws-iam

Claude Code

Copy this prompt.

Install the aws-iam skill in Claude Code. Run this command:

/plugin install aws-core@claude-plugins-official

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The plugin route installs aws-core: 25 skills (counted from plugins/aws-core/skills in the repo tree), an AWS MCP server entry in .mcp.json, and a PreToolUse hook that runs secret-safety.py. The skills routes install skill files only. The README says you need uv, and an AWS account with credentials set up locally for API calls and scripts. It needs no sibling skills.

From the project README. Source.

What's in the skill

When it loads

Provides verified corrections for IAM behaviors that AI agents frequently get wrong — policy evaluation edge cases, trust policy gotchas, STS session limits, Organizations quirks, and SAML/MFA specifics. Also provides structured workflows for IAM role management and baseline policy generation from application source code or a Terraform plan JSON. Covers condition operator safety (ForAnyValue/ForAllValues with Null checks), bucket policy deny patterns (VPC endpoint restrictions, org paths), confused deputy protection, and service role creation for AWS services (Glue, CloudTrail, Lambda, ECS, etc.) with aws:SourceAccount/aws:SourceArn trust conditions. Applies when creating IAM roles, writing IAM or bucket policies, generating policies from application source code or a Terraform plan JSON, working with STS, Organizations, or condition operators, or any task needing a service or execution role. Does not cover non-IAM authorization like Cognito user-pool policies or app-level RBAC.

Sections in SKILL.md

About This Skill
Common Workflows
Verified Edge Cases

Key facts

Author: Amazon Web Services (vendor's own repo; named in the aws-core plugin manifest)
License: Apache-2.0
Installs on skills.sh: 10,540 (Nov 2, 2026)
GitHub stars: 3,366 for the whole aws/agent-toolkit-for-aws repo (Nov 2, 2026)
Last repo commit: Oct 30, 2026
What it includes: 5 files including SKILL.md: 4 reference guides. No scripts
skills.sh audits: Gen Agent Trust Hub: pass; Socket: pass; Snyk: warn (Nov 2, 2026)
Needs: uv for the policy generation command. An AWS account with local credentials is needed for API calls, per the README
Part of: The aws-core plugin (version 1.2.0) in aws/agent-toolkit-for-aws: 25 skills counted from the plugin folder, an AWS MCP server entry and a PreToolUse hook
Works in: Claude Code, Codex, Cursor, Devin, Kiro and fx, per the README

What it includes

10 groups of IAM corrections, from CloudTrail and STS to SAML, MFA and SigV4
Examples include ForAllValues with a missing key, and iam:PassRole with a wildcard resource.
Verified edge cases
Create and maintain service and execution roles, with trust and confused deputy rules
Write · IAM role management
Proposes iam-policy-autopilot generate-policies for source code or a Terraform plan JSON
Python, Go, TypeScript/JavaScript and Java are the languages the guide lists.
Runs code · Policy generation from code
Checks uvx iam-policy-autopilot --version, and may try brew or pip installs
Runs code · Autopilot availability check
Reads env vars, runs aws sts get-caller-identity, checks project config files
Looks in terraform.tfvars, cdk.json, samconfig.toml, .env and serverless.yml.
Runs code · Account and region discovery
Looks up IAM actions in the Service Authorization Reference when Autopilot does not apply
Network · Policies from API operations
--upload-policies sends policies to IAM; the guide says only with explicit confirmation
Write · Upload generated policies
Wildcard resources, ForAnyValue and ForAllValues without Null checks, tag key case
Common pitfalls
Five patterns: actions per operation, action names, ARN formats, condition keys, listings
Network · Service Authorization queries

From the official docs

Permissions

Medium risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It runs a downloaded tool at its latest version
The guide uses uvx iam-policy-autopilot@latest, which fetches and runs the newest release each time. Review the command and the policy it produces before you apply anything. github.com
It can upload policies to IAM
The --upload-policies flag sends generated policies to IAM. The guide says never to upload or apply policies without explicit user confirmation, so refuse the flag unless you want it. github.com
The plugin route adds an AWS MCP server and a hook
aws-core's .mcp.json starts mcp-proxy-for-aws-cli 1.7.0 through uvx against https://aws-mcp.us-east-1.api.aws/mcp with a --skip-auth flag. Its hook runs secret-safety.py before Bash and AWS tool calls, and the script denies direct Secrets Manager value fetches. The script has a 5 second timeout, and its own comment says a hang would let the call through. github.com
It can install packages on your machine
If uvx is missing, the guide tries brew install uv or pip install uv, and then pip install iam-policy-autopilot. Say no if you do not want installs. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode in the skill. Use an AWS role with limited permissions.
Talks to	servicereference.us-east-1.amazonaws.com (Service Authorization Reference). docs.aws.amazon.com (SKILL.md tells Claude to fetch known documentation URLs). Package downloads through uvx, brew or pip. aws-mcp.us-east-1.api.aws, if you use the plugin route
Sign-in and keys	SKILL.md does not store credentials. It uses whatever AWS credentials are configured locally, and the policy guide tells Claude not to output config values other than the account ID and region.
Scopes	Claude can run uvx iam-policy-autopilot@latest (policy generation and a version check) If uvx is missing, the guide tries brew install uv or pip install uv, then pip install iam-policy-autopilot Claude can run aws sts get-caller-identity and aws configure get region, and read env vars and project config files for the account ID and region The Service Authorization Reference path uses curl against servicereference.us-east-1.amazonaws.com when the query tool is unavailable

Questions

How do I install the aws-iam skill?

Use the install box at the top of this page. The README documents the aws-core plugin for Claude Code and a skills CLI command for the whole repo. The box also shows a command for this skill only.

What does the aws-iam skill do?

It corrects IAM facts that AI agents often get wrong, and it gives Claude workflows for IAM roles and baseline policies. SKILL.md says it is not a comprehensive IAM guide and tells Claude to check specific limits and API names against AWS documentation.

Is this an official AWS skill?

Yes, it is in the vendor's own repo: AWS's aws/agent-toolkit-for-aws, and the aws-core plugin manifest names Amazon Web Services as author.

Does it write my IAM policy for me?

For source code in a supported language or a Terraform plan JSON, it tells Claude to propose the iam-policy-autopilot command and not to hand-write the policy. For other requests, Claude builds the policy from the Service Authorization Reference.

Does it need AWS credentials?

Some steps do. The README says local AWS credentials are needed for API calls and script execution, but not for documentation search or skill discovery. The policy guide runs aws sts get-caller-identity to find your account ID when it can.

Can it change my IAM setup?

The guide's Autopilot command can upload policies to IAM with --upload-policies, and it says never to upload or apply policies without explicit user confirmation. SKILL.md says to use the best available tool for AWS operations, and that the AWS MCP server is recommended but not required. Review each call.

Is it safe?

The skill bundles no scripts, but its guides tell Claude to run commands: uvx downloads and runs iam-policy-autopilot, and it may try brew or pip installs. The plugin route also adds an AWS MCP server and a hook. Read each command before you approve it.

Information

updated Nov 2, 2026

Repository: aws/agent-toolkit-for-aws
Folder: skills/core-skills/aws-iam

aws/agent-toolkit-for-aws skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 10

aws-iam SKILL.md · github.com
aws-iam-policy-generation.md · github.com
aws-iam-role-management.md · github.com
common-pitfalls.md · github.com
service-authorization.md · github.com
agent-toolkit-for-aws README · github.com
aws-core plugin.json · github.com
aws-core hooks.json · github.com
secret-safety.py · github.com
aws-core .mcp.json · github.com