Agentic BAS — breach & attack simulation for AI agents and MCP servers. Drives an attacker (curated seed payloads or any LLM via litellm) against a target, proves impact with per-run canaries (not an
Agentic BAS — breach & attack simulation for AI agents and MCP servers. Drive an attacker (curated seed payloads and/or any LLM) against a target, prove impact with per-run canaries, and get a report mapped to the OWASP LLM Top 10 and MITRE ATLAS.
INTENTIONALLY VULNERABLE. The fixture in fixture/ is for authorized, self-hosted testing only. Run it on localhost; never expose it to a network.
uvx aphasia-agentry --help run without installing pipx install aphasia-agentry or: pip install aphasia-agentry
git clone https://github.com/mddanish/Aphasia-Agentry cd Aphasia-Agentry/engine && uv run aphasia --help
Most agent red-teaming grades a model's output with an LLM judge. Aphasia Agentry asks a harder question — did the attack actually reach impact? — and answers it with evidence:
- Proof, not judgement. A finding is a touched canary record, a used honey token, or a mutating tool call captured by a record-only mirror. Never a model's self-assessment. - Any LLM, or none. Seed mode runs curated payloads with no LLM at all (reproducible, runs in seconds). Add an adaptive attacker via litellm: OpenAI, Anthropic, Gemini, Groq, OpenRouter, Bedrock, local vLLM/LM Studio, Ollama. - Full OWASP LLM Top 10. 11 scenarios across LLM01–LLM10, mapped to MITRE ATLAS. - Safe by construction. The bundled vulnerable fixture is sandbox-inert — real exploit, fake blast radius: no disk, exec, env or network from any attack.
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/mddanish--Aphasia-Agentry.html)
Stop your AI from making things up — it proposes, deterministic tools decide, every claim checked against ground truth with evidence. Grounded facts and context survive resets. Reverse engineering is
zhuyansen/new-term-radarDaily radar for new AI terms: names of new models, products and repos, caught on X, Hugging Face, GitHub and sitemaps before keyword tools list them.
yzg1226/mcd-party-officer🍟 麦当劳门店派对成团调度 Skill — 把派对从「策划」推进到「办得成」：成团概率雷达 / 包场vs拼团 / 择日黄历 / 儿童营养指数 / 招募卡与邀请函
fellowgeek/mcp-memoryAn OKF-backed Model Context Protocol (MCP) server delivering persistent long-term memory and SQLite FTS5 search for AI agents.
hirotomasato/yowesGenerate realistic teacher documents (ID cards, licenses, letters) for 13 countries via MCP.
hujjun/mcd-saver-strategist麦当劳省钱点餐助手（麦门精算师）｜麦当劳优惠券组合优化 Skill：拉取麦当劳菜单与全部可用优惠券，算出最省的点餐组合，巨无霸套餐 ¥47 → ¥21.9（省 53%）。McDonalds MCP（麦当劳 MCP）coupon optimizer · McDonalds China MCP Server · Agent Skill for WorkBuddy · 麦当劳程序员创意开发大赛参赛作品
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.