Guide · Agent skill
How to use the Initial Access & Phishing agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
This guide explains the initial-access-phishing agent skill from the CyberStrikeAI project, a short reference card that teaches an AI agent the common ways attackers get a first foothold during an authorised penetration test. It is written for red-team operators, security teams and students who may never have used an agent skill before. By the end you will know what the skill contains, how to install it, how to phrase requests that trigger it, and where its limits and safety requirements are.
Initial Access & PhishingAIPentest/CyberStrikeAI · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Get the initial-access-phishing skill ZIP from the directory page.
- 2Unzip into the skills folder. Unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the ZIP.
- 4State your authorised scope. Describe the authorised engagement and the initial access technique you want to plan.
- 5Review the output. Check the agent's response against your engagement scope and rules of engagement before acting on it.
What is the initial-access-phishing skill?
An agent skill is a folder containing a SKILL.md file that gives an AI agent extra context for a specific kind of task. This skill covers initial access, meaning the first step by which an attacker moves from outside an organisation to inside it, and it is aimed at work done before a penetration test's first foothold.
The file is a reference card rather than a procedure. It names techniques and the tools associated with them, and it gives little step-by-step detail.
- Credential spraying: trying weak or common passwords across all accounts, with 1-2 attempts per account to avoid lockouts, using leaked data, email format guesses or default credentials as sources.
- AiTM phishing: using reverse-proxy tools such as evilginx3 or Modlishka to relay a real login site and capture session cookies and tokens, which can bypass MFA.
- Device code phishing: abusing the Microsoft device code flow to get the victim to enter a code, yielding access and refresh tokens. Tools named include TokenTactics and AADInternals.
- Phishing payloads and delivery: gophish for sending mail and hosting landing pages, plus payload types such as lnk, iso, macros, HTA and OneNote, and gateway evasion methods.
- OAuth consent phishing: tricking users into granting a malicious app excessive scopes such as Mail.Read or offline_access for long-term access.
- Social engineering: OSINT on organisation charts, suppliers and staff habits, pretexts such as IT support, suppliers or HR, and vishing including deepfake voices.
How do I install the skill?
You can install the skill in one of two ways. Download the skill ZIP from the directory page and unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want it available only for that project.
If you use the Claude apps instead, upload the ZIP under Settings → Capabilities → Skills.
- Personal install: unzip into ~/.claude/skills/.
- Project install: unzip into the project's .claude/skills/ folder.
- App install: upload the ZIP under Settings → Capabilities → Skills.
How do I trigger the skill with a natural request?
The skill's description tells the agent when to use it: initial access, phishing, AiTM, device code, or social engineering. You do not need a special command. Name the situation and the technique in plain language.
Include the authorisation context. The skill is intended for authorised engagements, and a request that states the scope makes that clear to both you and the agent.
- Example: "I have written authorisation for an external test of our Microsoft 365 tenant. Explain how device code phishing works so I can plan the test."
- Example: "Help me outline the initial access options to check in a phishing-resistance review of our MFA setup."
- Example: "What social engineering pretexts should my red team consider for this engagement, and how do I document them?"
What does the skill need and how does it run?
The skill lists no requirements. It does not run scripts and does not use the network. It is a single file, SKILL.md, that the agent reads as context.
Because it runs nothing itself, its output depends on what the agent and your other tools do. The file points to two related skills, redteam-opsec for avoiding detection after entry and capability-primitive-search for chaining capabilities.
- Requires: nothing listed.
- Runs scripts: no.
- Uses network: no.
- Files: SKILL.md only.
What safety rules should I follow?
The content is an offensive playbook. Use it only against systems and people you have written authorisation to test. Attacking without authorisation is illegal, and the skill itself does not add any consent or scope checks.
Because the skill does not run scripts or touch the network, the risk lies in what you do with its guidance. Keep your engagement scope, rules of engagement and written permission in hand for every action.
- Test only with written authorisation and a defined scope.
- Do not use the skill for step-by-step attacks on people or systems you do not own or have permission to test.
- Keep records of actions taken during the engagement.
Tips for getting useful results
Ask for the reasoning behind each technique and how defenders would spot it, since the file focuses on offence. For defensive guidance such as detection or hardening, pair this skill with other material, because the file does not provide it.
Be specific about your target environment, such as whether the organisation uses Microsoft 365, so the agent can focus on relevant techniques.
- State the target platform and authorised scope in your first message.
- Ask for a structured plan or checklist rather than exact commands, since the source gives little procedural detail.
- Use the related skills named in the file when you move from entry to stealth or chaining.
What people use it for
- Planning the initial access phase of an authorised red-team engagement.
- Reviewing which phishing and MFA-bypass techniques a security team should test against.
- Learning how initial access attacks are structured for a security course.
- Building a checklist of social engineering pretexts for an approved assessment.
Frequently asked questions
How do I install the initial-access-phishing skill?
Download the skill ZIP from the directory page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
Does the initial-access-phishing skill require anything else?
The source lists no requirements. The skill does not run scripts and does not use the network, and it consists of a single SKILL.md file.
Is the initial-access-phishing skill free?
The skill is licensed under Apache-2.0. The source does not state pricing beyond that licence, so check the licence terms before using or redistributing it.
What is the difference between AiTM and device code phishing?
AiTM phishing uses a reverse-proxy tool such as evilginx3 or Modlishka to relay a real login page and capture session cookies. Device code phishing tricks the victim into entering a code so the attacker receives access and refresh tokens.
Can this skill replace a penetration testing tool?
No. The skill is a reference card that names techniques and tools. It does not run them, and the source gives little procedural detail.
Who should not use this skill?
Anyone without written authorisation to test the target, and beginners looking for step-by-step attack instructions. Defenders needing detection or hardening guidance should look elsewhere, since the file describes offence only.
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 red-team operators, penetration testers preparing an authorised engagement, security teams validating MFA and phishing resistance, and students learning how initial access attacks are structured. It is not suitable for attacking anyone without written authorisation, for beginners wanting step-by-step instructions, or for defenders who need detection or hardening guidance.
Download the skill