Guide · Agent skill
How to use the Offensive Cloud Security Methodology agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The offensive-cloud skill is a written attack playbook that an AI agent reads before it works on cloud accounts in AWS, Azure or GCP. It is meant for authorized security tests, where the agent helps you identify the current identity, map its permissions, and document an attack chain. This guide explains what the skill changes, how to install it, how to ask for work in plain language, and what you need in place before you run it.
Offensive Cloud Security MethodologySnailSploit/Claude-Red · MIT Download the skill Quick start
- 1Download the ZIP. Get the offensive-cloud ZIP file from this page.
- 2Unzip into a skills folder. Extract the offensive-cloud folder into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. Alternatively, open Settings, go to Capabilities, then Skills, and upload the skill there.
- 4Install the required tools. Install the cloud CLI for your target (AWS CLI, Azure CLI or gcloud CLI) plus Pacu, ScoutSuite, Prowler, ROADtools and gcp_enum, following each project's own instructions.
- 5Confirm scope and credentials. Make sure you have valid credentials or a test account and written authorization that covers the cloud accounts you will test.
- 6Ask in plain language. Describe the cloud, the identity you hold and the goal, starting with read-only discovery.
- 7Review and document. Check each proposed step before approving it and keep a log of identities, resources and timestamps for the report.
What the offensive-cloud skill makes the agent do
An agent skill is a folder containing a SKILL.md file with instructions. When a skill matches your request, the agent reads those instructions and follows them. This one covers cloud security testing across AWS, Azure and GCP and is written from an attacker's point of view rather than as hardening advice.
The skill's workflow is: identify the cloud and the identity you have, enumerate without making changes, map permissions to known privilege-escalation patterns, locate data and persistence options, and record the chain of events for a report. Its content covers credential discovery, enumeration with tools such as Pacu, ScoutSuite, Prowler, ROADtools and gcp_enum, and known IAM misconfigurations.
- Confirms who the credentials belong to, starting with read-only identity checks
- Suggests the enumeration tools relevant to the cloud being tested
- Matches permissions against known privilege-escalation paths
- Helps structure findings and timestamps for the written report
Install the offensive-cloud skill
Download the ZIP file for the skill from this page. Unzip it so that the offensive-cloud folder, including its SKILL.md file, sits in your personal skills directory at ~/.claude/skills/. To limit the skill to one project, place it in that project's .claude/skills/ folder instead.
If you use the Claude apps, you can instead upload the skill under Settings, then Capabilities, then Skills. Use whichever route fits how you work; the skill content is the same either way.
- Personal install: ~/.claude/skills/ (available in every project)
- Project install: the .claude/skills/ folder inside a specific project
- App install: Settings, then Capabilities, then Skills, then upload
Install the tools the skill depends on
The skill does not bundle its tools. Depending on which cloud you are testing, install the AWS CLI, the Azure CLI (az) or the gcloud CLI. The skill also relies on Pacu, ScoutSuite, Prowler, ROADtools (roadrecon) and gcp_enum, which you must install separately. Follow each project's own installation instructions, because this page does not provide install commands for them.
You also need valid cloud credentials or a test account that your written authorization covers. The skill does not supply credentials.
- AWS CLI, Azure CLI (az) or gcloud CLI, depending on the target cloud
- Pacu, ScoutSuite, Prowler, ROADtools (roadrecon) and gcp_enum
- Valid cloud credentials or an authorized test account
How to trigger the skill with natural requests
You do not need a special command. Describe the task in plain language and name the cloud. The skill is designed to apply when the engagement scope is cloud accounts, when you already hold cloud credentials, or when you are assessing cloud posture. Mention the scope and the authorization so the agent works within it.
Good requests state the cloud, the identity you have, and the goal. Start with read-only discovery, then ask for the next step once you have reviewed the results.
- Within our signed scope for account 123 test environment, tell me which identity these AWS credentials represent and what policies are attached to it.
- I have an authorized Azure lab tenant. Show me my current subscription and role assignments before we look at anything else.
- Using the GCP credentials for our approved test project, list the auth accounts available and summarise what they can reach.
- Review the findings so far and draft the attack chain with timestamps, identities and resources for the report.
Work through an authorized test step by step
Keep each step small so you can check the output before moving on. The skill's own workflow favours enumeration without writes, so review what the agent proposes before approving any action that changes a resource. Ask the agent to explain why a step is needed and what it would change.
Record each identity, resource and action as you go. The skill recommends documenting the kill chain for the report, and that record is easier to build during the test than afterward.
- Confirm scope and authorization before any action
- Start with read-only identity and permission checks
- Approve each change individually and note what it altered
- Keep a running log of identities, resources and timestamps
Safety notes: scripts, network and production systems
According to its listing, this skill does not run scripts itself, but it does use the network, because the tools it directs the agent to use contact cloud APIs and metadata endpoints. Everything it does should be limited to systems you own or have written permission to test.
Some steps described in the skill, such as sharing snapshots, creating keys, or changing logging, alter real resources and can expose data or cause outages. Do not run these against production without a change-control plan agreed in advance. The skill is not suitable for hardening an account; it is for attacking under authorization.
- Use only with written authorization covering the target cloud accounts
- Prefer a lab or dedicated test account over production
- Require change control before any step that alters real resources
- Expect the skill to describe offensive techniques, not defensive fixes
What people use it for
- Penetration testers checking which permissions a cloud identity holds during a scoped engagement
- Red team members practicing cloud attack paths in a lab environment
- Security engineers studying how attackers chain IAM permissions across AWS, Azure and GCP
- Assessors documenting an attack chain with identities and resources for a written report
Frequently asked questions
Is the offensive-cloud skill free to use?
The skill is published under the MIT licence, as stated in its listing. The tools it relies on, such as Pacu, ScoutSuite, Prowler and ROADtools, are separate projects with their own licences, which you should check before use.
What do I need before installing the offensive-cloud skill?
You need an agent environment that supports skills, the ZIP file from this page, and the cloud CLI for your target. You also need Pacu, ScoutSuite, Prowler, ROADtools (roadrecon) and gcp_enum installed separately, plus valid credentials or an authorized test account.
How do I install an agent skill?
Unzip the skill folder into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings, then Capabilities, then Skills.
Does the skill run scripts or use the network?
According to its listing, the skill does not run scripts, but it does use the network, since the tools it directs the agent to use contact cloud APIs and metadata endpoints. Use it only against systems you are authorized to test.
Can I use this skill to harden my own cloud account?
No. The skill is written as offensive tradecraft and does not give defensive hardening guidance. For securing an account, look for a defensive posture or hardening resource instead.
What is the difference between this skill and tools like Prowler or ScoutSuite?
Prowler and ScoutSuite are enumeration tools that the skill tells the agent to use. The skill itself is the methodology that guides the agent in choosing those tools, mapping permissions to escalation paths and organizing findings into an attack chain.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits penetration testers and red teamers who have written authorization for cloud accounts and want an agent that follows a structured AWS, Azure and GCP attack playbook. It is a poor fit for anyone without authorization, for production work without change control, or for people who want cloud hardening advice.
Download the skill