red-team plugins

by hypnguyen1209

Spec-driven offensive-security framework for Claude Code: 31 kill-chain skills, executable safety controls (scope/finding/OPSEC discipline), pattern-learning memory, and a bounded engagement engine — with a SessionStart dispatcher that enforces skill-invocation discipline.

by Ross Young

Agentic-Security: SAST + SCA + secrets + IaC + LLM-sec scanner with deterministic fix toolchain. Exposes a Model Context Protocol (MCP) server with 17 tools that Codex CLI can call directly.

Agent configuration utilities - project assimilation, config auditing, teammate definitions, MCP management, and hooks configuration

by Adam Rhys Heaton (Ap6pack)

Claude-native authorized external recon and ASM methodology bundle with 90 documented capabilities across 11 skills, deterministic Python controls, an optional loopback-only limited-control web plane, explicit human-reviewed finding promotion, and optional fixed policy-gated MCP enrichment. It does not provide unrestricted exploitation or autonomous recon orchestration.

by vikash ruhil

Loomwright is a plan-first, multi-agent development system for Claude Code. Launch Pad turns a goal into a reviewed brief; Supervisor executes it autonomously across parallel git worktrees with self-heal; dedicated agents cover code review, red-teaming, QA, and standalone PR review-and-heal. 14 agent roles, 24 slash commands, 42 skills, 39 quality gate hooks. Full history in CHANGELOG.md.

by Stellars Henson

Critical document analysis with persona-driven risk scoring, iterative scorecard evaluation, and versioned corrections. Includes adversarial-review - hostile red-team review of code and artefacts via fresh claude -p reviewers with eleven pluggable expert adversaries.

by Gregory Block

The research debate engine for Special Circumstances: an additive blue side builds, a subtractive red side audits and owns the pass/fail gate, an engine runs the mechanics and the bench rules the final compromise. Best-of-N, the Catechism (a worth-our-time decision), and the full record preserved.

by XRenSiu

Agent Teams Product Development Pipeline - 7-phase adversarial collaborative development with multi-agent debate, competition, and cross-validation at every critical decision point, including red team attacks, adversarial debugging, independent bug fix loop, and requirement verification.

by DBHQ

Put a second agent on the work - as an adversary or as a partner - and get back a result you can cite. Five named patterns: red-team attacks an idea without being shown your evidence, second-opinion judges your material without being shown your conclusion, verify rules on finished work against stated constraints, collaborate is a peer conversation, and debate runs blind proposals into adversarial rounds. Runs on Codex, opencode or GitHub Copilot behind one provider layer, and records what each run was told and what it was not.

by raystyle

rust-script 渗透测试体系:五域(资产侦察/漏洞扫描/漏洞利用/后渗透操作/攻击产物分析)单文件脚本、pentest-lib 框架、wssd 远程壳 daemon、vulhub 实证回归。仅限合法授权测试。