Cross-model orchestration for Claude Code — Claude builds, Codex validates. Blind TDD, adversarial stress testing, mixed-model teams, and automatic fallback. Two AI models enter, better code leaves.
Adversarial MCP server benchmark suite for testing tool-calling security, drift detection, and proxy defenses
https://bun.com/blog/bun-in-rust turn this into a skill for an agents.
autodev — LLM-driven autonomous software development loop with YAML state machine, TwoRoundGate adversarial quality gates, and three-tier human intervention (auto/HITL/HOTL)
Deterministic audit layer for GPX tracks - makes timestamp, sampling and motion anomalies explicit as schema-versioned JSON. Observes, never corrects.
Multi-persona deep-experience critique simulator: 5-7 AI-played personas actually run your code / pages / long-form text, then deliver a script-verified newspaper-style critical review. Zero npm deps, Node 22+.
Stop taking your coding agent's word for it: script-computed verdicts, re-executable reports (zero LLM calls), N-version tournament delivery, CI evidence gate. Claude Code plugin + 17 tool adapters.
Adversarial JWT library test harness: 8 libraries, 4 ecosystems, RFC 7515 compliance.
Adversarial Multi-Model Code Review — pit multiple LLMs against your code simultaneously, cross-reference their findings, and interactively triage the results.
To associate your repository with the adversarial-testing topic, visit your repo's landing page and select "manage topics."