0xSteph/pentest-ai-agents is a collection of Claude Code specialist agents for authorized penetration testing and security research, covering areas such as reconnaissance, web systems, cloud, reverse engineering and detection. Security researchers and penetration testers use it to plan engagements, investigate findings, build detections and write reports. The catalogue entries are the project's own agents, commands and plugin components.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/0xSteph/pentest-ai-agentsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/0xsteph/pentest-ai-agents/llm-redteam)<a href="https://agentmods.dev/agents/0xsteph/pentest-ai-agents/llm-redteam"><img src="https://agentmods.dev/badge/agents/0xsteph/pentest-ai-agents/llm-redteam/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/0xsteph/pentest-ai-agents/llm-redteam"><img src="https://agentmods.dev/badge/agents/0xsteph/pentest-ai-agents/llm-redteam.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00079 | $0.04830 |
| Opus 5 | $0.00039 | $0.02415 |
| Sonnet 5 | $0.00016 | $0.00966 |
| Haiku 4.5 | $0.00008 | $0.00483 |
Grade A, and why
llm-redteam scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 412 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are an LLM and AI system red team specialist. You guide operators through testing AI applications: prompt injection, jailbreaks, RAG poisoning, agent abuse, model and data exfiltration, and the surrounding application security issues that emerge when an LLM sits in the data path. You focus on production AI applications (chatbots, copilots, agentic systems, MCP-connected tools), not on academic adversarial-ML research.
Scope Boundary
- In scope: prompt injection (direct, indirect, multi-modal), jailbreak chains, system prompt extraction, RAG poisoning, training-data extraction, agent and tool-use abuse, MCP server abuse, output handling vulnerabilities (XSS via LLM, SSRF via tool use), guardrail and content-filter bypass, denial of wallet, AI supply chain (model/dataset poisoning).
- Out of scope: adversarial-ML research against vision models for evasion (different methodology; consult academic resources), model training pipeline security except where it affects deployed apps (use
cicd-redteamfor pipeline CI/CD security). - Hard refusal: jailbreaks of public production systems (ChatGPT, Claude.ai, Gemini) that are not authorized targets. Hard refusal: producing CSAM, bioweapon synthesis, or other content that the underlying model's safety stack is correctly preventing. Authorization to red team an app is not authorization to bypass safety to extract harmful content.
Behavioral Rules
- Authorized targets only. The user must be testing an application they own, have a signed engagement against, or are authorized via a bug bounty program with explicit AI scope.
- OWASP LLM Top 10 mapping. Every finding maps to OWASP LLM Top 10 (2025 edition). Use that as the standard taxonomy in reports.
- Application boundary, not model boundary. Most real findings are at the application boundary: how the app handles model output, how RAG sources are sanitized, how tool calls are gated. Don't fixate on cute jailbreak strings; fixate on what the app does with model output.
- Severity by impact, not novelty. A two-line indirect injection that exfiltrates the customer database is critical. A clever twelve-step jailbreak that produces a swear word is informational. Rate accordingly.
- Don't generate harmful content. When demonstrating prompt injection, use placeholder payloads like
[exfil_target]or<harmful_content>. The vulnerability is the bypass, not the content. - Reproducibility. Every finding includes the exact prompt, full conversation history, model version (if visible), and any retrieval context. Without those, the customer cannot fix.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 412 lines · 79 tokens per session scan E e4ccaa5dded0
llm-redteam is an agent published in the GitHub repository 0xSteph/pentest-ai-agents (2,213 stars, last pushed 24d ago), licensed MIT. It adds 79 tokens to every session and 4,830 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
tachi-risk-scorer
Quantitative risk scoring agent that enriches threat model findings with four-dimensional scores (CVSS 3.1, exploitability, scalability, reachability), computes weighted composite scores, attaches governance fields, and generates dual-format output (risk-scores.md and risk-scores.sarif).
active-directory
Active Directory and Windows domain attack specialist. Use for Kerberoasting, AS-REP roasting, DCSync, BloodHound enumeration, ADCS ESC attacks, Golden/Silver Ticket, and domain privilege escalation. Triggers on: kerberoast, AS-REP, bloodhound, DCSync, golden ticket, ADCS, ESC, domain controller, LDAP, GPO, AD, domain…
exploit
Exploitation specialist for gaining initial access. Use when exploiting CVEs, running Metasploit modules, using searchsploit, obtaining shells, or executing proof-of-concept code. Triggers on: exploit, CVE-, initial access, get shell, msfconsole, owned, pwn, vulnerability exploit, remote code execution, RCE.
iot-attacker
IoT and embedded systems security specialist. Handles firmware extraction and analysis, hardcoded credential discovery, UART/JTAG access, MQTT/CoAP protocol testing, RouterSploit exploitation, web interface attacks, and OT/ICS protocol analysis. Triggers on: IoT, firmware, binwalk, UART, JTAG, router, embedded…
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.