A large collection of Claude Code subagents for offensive security research, with 50 agents, install options, and a plugin-based workflow. Latest release in v3.2.0 (2026-05-03) adds 4 new agents, bringing total to 50.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
pentest-ai-agents is a collection of 50 Claude Code subagents designed to function as an offensive security research assistant. The agents cover recon, web, AD, cloud, mobile, wireless, social engineering, payload crafting, reverse engineering, detection engineering, forensics, exploit chaining, reporting, and more. Install the agent files, open Claude Code, and Claude routes tasks to the appropriate specialist automatically. No servers, no Python dependencies, and no setup beyond copying files.
How it works
The project provides a set of specialized agents that map to offensive-security domains. Some agents have Tier 2 capabilities, meaning they can execute tools directly when given scope. The README describes an agent map showing how tasks flow from engagement planning through reconnaissance, exploitation, post-exploitation, and reporting, with specific agents responsible for each step. There is a Cheatsheet with commands and workflows, including /recommend, /agents-for, and db/doctor.sh for auditing installed CLI tools.
Getting started
Install options include:
curl -fsSL https://raw.githubusercontent.com/0xSteph/pentest-ai-agents/main/install.sh | bash
This clones the repo to a temp dir, copies the agents to ~/.claude/agents/, and exits. Idempotent: safe to re-run for updates.
Or install as a Claude Code plugin:
/plugin marketplace add 0xSteph/pentest-ai-agents
/plugin install pentest-ai-agents@pentest-ai-agents
Then open Claude Code and issue a task like:
"Plan an internal network pentest for a 500-endpoint AD environment with a 2-week window."
Claude routes to the engagement planner agent and produces a phased plan with MITRE ATT&CK mappings.
Other install options are listed, including cloning the repository and running install.sh with various flags. See INSTALL.md for more details.
Recent releases
Latest release is v3.2.0 (2026-05-03):
- 4 new agents (c2-operator, container-breakout, opsec-anonymizer, llm-redteam) increasing total to 35 at that time.
- Improvements include a tightened scope guard, findings DB v2, and an agent map diagram.
The README also documents v3.1 and v3.3 updates, including 15 new agents (for v3.3) and installer improvements, CI validation, and offline Docker bundle notes.
Traction
Stars: 2072. Forks: 400. Open issues: 3.
Behind the repo
Not provided in the portion available. The project describes a plugin-based installation and agent roster, but there is no explicit "Behind the repo" section in the provided content.
Caveats
License: MIT. Created: 2026-03-28. Last push: 2026-06-22. Language: Shell. The README indicates install.sh options (e.g., --tools, --global, --lite) and that there are CLI tools installable via install.sh. There is a note about a minimal offline Docker bundle in v3.3. No further licensing or open issues details are presented beyond the overall license and counts.