Pentestizer

agent
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

A reusable AI skill for authorized penetration testing, covering reconnaissance, vulnerability assessment, security analysis, validation, and reporting.

README.md

Pentestizer icon

Pentestizer

Practical, modular penetration-testing skills for AI agents and security practitioners.

Validation MIT License Agent Skills

Install · Quick start · Skills · Documentation

Pentestizer is an open collection of focused Agent Skills for planning and performing authorized security assessments. Each skill provides a repeatable workflow, practical tool guidance, verification steps, and reporting expectations.

The project is designed to be usable by AI coding agents and human penetration testers. It is not a scanner, exploit framework, or substitute for professional judgment.

Authorization first. Only test systems for which you have explicit, current written authorization. A public hostname, a prior permission, or the existence of a bug bounty program does not by itself authorize testing.

Install

Use with an Agent Skills-compatible client

Pentestizer follows the open Agent Skills directory convention. You can point your agent at this repository's skills/ directory or copy the skill folders you need into the skills directory supported by your client.

If your client supports the Skills CLI, you can also install directly from GitHub:

npx skills add asimawdah/Pentestizer

Installation behavior depends on your client/CLI support for multi-skill repositories. For manual placement and client-specific notes, see Getting Started.

Included skills

Skill Purpose Common tools
Engagement scoping Authorization, rules of engagement, scope and stop conditions Scope document, scope_guard.py
Passive reconnaissance Organize passive asset discovery and verify scope before probing Subfinder, assetfinder, dnsx, httpx
Network service enumeration Inventory explicitly authorized hosts and exposed services Nmap
Web application testing Structured web testing based on OWASP WSTG Burp Suite, OWASP ZAP, browser devtools, curl
API security testing Assess API authentication, authorization, input handling and configuration Burp Suite, OWASP ZAP, Postman, curl
Vulnerability assessment Carefully scoped detection of known issues and exposures Nuclei
Exploit research Find and verify advisory/PoC provenance and applicability CVE, NVD, OSV, Exploit-DB, SearchSploit
Exploit PoC development Build minimal, non-destructive proof-of-concepts Python, pwntools, GDB, Burp Suite
Exploit adaptation review Inspect and constrain third-party PoCs before use Source review, hashes, test harnesses
Exploit lab validation Validate reviewed PoCs in isolated, disposable labs Metasploit, Docker, VirtualBox, Ghidra
TLS configuration review Review TLS versions, ciphers, certificates and transport security testssl.sh, SSLyze, OpenSSL
Source code security review Static analysis, dependency and secret checks Semgrep, Gitleaks, Trivy
Security reporting Coordinate the complete report lifecycle Canonical report dataset
Report authoring Write executive summaries and validated technical findings Report schema, finding template
Multi-format export Export Markdown, HTML, JSON, CSV, SARIF, DOCX and PDF report_builder.py, Pandoc
Evidence redaction Prepare safe evidence and remove sensitive data/metadata ExifTool, MAT2, PDF redaction tools
Report quality assurance Review scope, severity, evidence and generated documents QA checklist, SARIF validation
Android application security Android platform and app binary assessment MASVS/MASTG, MobSF, JADX, Frida
iOS application security iOS platform, Keychain and app binary assessment MASVS/MASTG, MobSF, Ghidra, Frida
Cross-platform mobile security Flutter, React Native and MAUI app assessment DevTools, Semgrep, MobSF
Desktop application security Windows, macOS and Linux thick-client assessment Sysinternals, Objective-See, Ghidra
Database security Relational and NoSQL database posture Vendor audit facilities, pgAudit, Trivy
CI/CD and supply-chain security Repository, workflow, runner, SBOM and artifact integrity zizmor, Scorecard, Syft, Grype, Cosign
AI, LLM and agent security LLM, RAG, agent, tool and MCP security Promptfoo, Garak, PyRIT, OWASP GenAI
Windows host security Windows baseline, patch, endpoint protection and host hardening Microsoft SCT, CIS Benchmarks
Active Directory security AD DS, Kerberos, LDAP, GPO, trusts and AD CS PingCastle, BloodHound CE, Certipy
Linux host security Linux identity, SSH, sudo, services, kernel and logging Lynis, OpenSCAP, osquery
Cloud security AWS, Azure and Google Cloud posture Prowler, ScoutSuite, provider-native tools
Container security OCI images, SBOM, runtime and registry controls Trivy, Syft, Grype, Dockle
Kubernetes security Cluster, RBAC, workloads, nodes and network policy kube-bench, Kubescape, kubeaudit
Network infrastructure security Firewalls, routers, VPN, segmentation and management plane Nmap, Wireshark, Zeek, Batfish
Identity and federation security Entra ID, Okta, Keycloak, OAuth/OIDC, SAML and SCIM Provider audit APIs, protocol specifications
Virtualization platform security VMware, Hyper-V, Proxmox and KVM/libvirt CIS Benchmarks, vendor audit APIs
Wireless security Wi-Fi, Bluetooth/BLE and wireless configuration Kismet, Wireshark, BlueZ
IoT and firmware security Firmware provenance, update security, debug interfaces and emulation EMBA, Binwalk, FACT, QEMU
ICS/OT security SCADA, PLC, HMI, industrial segmentation and safety GRASSMARLIN, OpenPLC, MiniCPS
Advanced reverse engineering Binary analysis, debugging and patch comparison Ghidra, Rizin, Cutter, radare2
Security lab management Isolated lab lifecycle, snapshots, repeatability and teardown Docker/Podman, QEMU, VirtualBox
Evidence and retesting Evidence integrity, validation, redaction and retest lifecycle SHA-256, report QA, retest records

Quick start

1. Install and prepare

Use one of the install options above, then review the safety requirements before any testing work:

  1. Security and authorization policy
  2. Engagement scoping skill
  3. Rules of engagement template

2. Define an engagement scope

Copy the example and replace its reserved example values with the exact assets and boundaries approved by the system owner:

cp examples/engagement.example.json engagement.json

Check a proposed target before using it with a testing tool:

python scripts/scope_guard.py --scope engagement.json --target https://app.example.test

The legacy scripts/scope_guard.py checks a host/IP against an offline allowlist. The Phase 1 pentestizer.scope engine additionally evaluates exclusions, protocol/port limits, URL path prefixes and timezone-aware authorization windows. Neither verifies the authenticity of owner authorization, resolves DNS, pins destination IPs, validates redirects, or replaces network egress controls. Validate scope with python scripts/validate_scope.py --scope engagement.json and follow the assessment engine guide.

3. Choose a workflow

Start with Getting Started, then select the relevant skill from the complete coverage matrix. Phase guides: Foundation and engine, Enterprise security, Applications, data and AI, and Advanced security and labs. Reporting details are in Report Generation and exploit workflows in Exploit Research and Validation.

Repository structure

Pentestizer/
├── skills/                 # 39 portable Agent Skills
├── pentestizer/            # Scope, registry, runner and result-normalization library
├── tools/                  # Reviewed tool registry and schema
├── coverage/               # Machine-readable assessment coverage catalogs
├── docs/                   # Phase guides, architecture and tool guidance
├── examples/               # Safe, fictional configuration examples
├── templates/              # Scope, rules of engagement, lab and report schemas
├── scripts/                # Validation, tool inventory, runner and reporting CLIs
├── tests/                  # Standard-library unit tests
└── .github/workflows/      # Automated validation

Design principles

  • Authorization first: no testing workflow starts before written scope and rules of engagement are established.
  • Least impact: prefer passive discovery, read-only review, low request rates and non-destructive verification.
  • Human oversight: require an operator to approve scope, test intensity, credentials, and any action with meaningful side effects.
  • Evidence over assumptions: distinguish confirmed findings from indicators and tool output that needs manual validation.
  • Reproducibility: record commands, tool versions, timestamps, target identifiers, and test conditions.
  • Data minimization: avoid collecting real user data, redact secrets, and securely handle assessment artifacts.
  • Portable format: skills follow the open Agent Skills directory convention and avoid mandatory third-party Python dependencies.

Report generation

Pentestizer includes a local report builder. Export the same reviewed JSON dataset to Markdown, HTML, JSON, CSV or SARIF without additional Python packages. DOCX and PDF exports use optional Pandoc support.

python scripts/report_builder.py --input examples/report.example.json --format html --output out/demo-report.html
python scripts/report_builder.py --input examples/report.example.json --format pdf --output out/demo-report.pdf --pdf-engine tectonic

The example data is fictional and must not be used as a real assessment. See Report Generation for the canonical schema, DOCX/PDF setup, report platforms and release checks.

Tooling and assessment engine

Pentestizer maintains a reviewed tool registry, an offline scope policy engine, a dry-run-first assessment runner, and adapters for normalizing selected scanner output. Tool availability checks do not install tools. Active execution is available only through reviewed profiles and explicit operator confirmation. See Tooling, Tool Management, and Assessment Engine.

Validation

The scope and report builders, validators, and tests use only the Python standard library. Pandoc is optional and required only for DOCX/PDF export:

python scripts/validate_skills.py
python scripts/validate_coverage.py
python scripts/validate_scope.py --scope examples/engagement.example.json
python -m unittest discover -s tests -v

References

The workflows are informed by established security guidance, including:

External standards and tool behavior can change. Check upstream documentation before an engagement.

Contributing

Contributions are welcome. Read CONTRIBUTING.md, keep each skill narrowly scoped, and include validation evidence for new workflows.

License

Pentestizer is released under the MIT License. External tools and referenced standards remain subject to their own licenses and terms.

Reviews (0)

No results found