AI repos for security engineers
Open-source AI projects worth it for people who secure systems, do pentesting, DevSecOps and AI red teaming, each with what it does for you. 14 repos, most relevant first.
-
rea
Let an agent inspect binaries, Electron apps and .NET assemblies via Ghidra, IDA or Hopper and return evidence.
-
strix
Autonomous agents pentest web apps and APIs and confirm each finding with a working proof-of-concept exploit.
-
Anthropic-Cybersecurity-Skills
800+ agent playbooks for threat hunting, forensics and red teaming mapped to MITRE ATT&CK and NIST.
-
SkillSpector
Scans agent skills and MCP tools for prompt injection and exfiltration, with SARIF output for CI gating.
-
android-reverse-engineering-skill
Decompile APKs, extract HTTP endpoints and auth patterns, and triage suspicious apps with Claude Code.
-
NemoClaw
Run agents inside OpenShell sandboxes with network policy to limit what an agent can reach.
-
ECC
AgentShield scans agent hooks, MCP config and permissions for risky settings.
-
gstack
Includes a security audit role to review changes with Claude Code.
-
system_prompts_leaks
Study leaked prompts to understand model guardrails and prompt-extraction exposure for AI red teaming.
-
CloakBrowser
Patched Chromium with fingerprint changes to study bot detection and test defenses on authorized sites.
-
CubeSandbox
Isolates untrusted LLM-generated code in separate VMs instead of on host machines.
-
camoufox
Study browser fingerprinting and test how bot-detection defenses respond to a spoofed Firefox.
-
peerd
Review a no-backend agent design with documented sandboxing and security model.
-
claude-code-mastery-project-starter-kit
Study hooks and rules that guard against secrets and unsafe commits in agent workflows.