meta-ai-support-prompt

Системная подсказка ассистента поддержки Meta AI Извлеченная системная подсказка от ассистента поддержки Meta AI от 1 июня 2026 года. Файлы system-prompt.md — Извлеченная системная подсказка ⚠️ Отказ от ответственности и юридическое уведомление Цель Этот репозиторий опубликован строго для...

CVE-2025-51860

CVE-2025-51860 취약점 설명 TelegAI는 AI 캐릭터를 구축하고 대화할 수 있는 웹 애플리케이션으로, 채팅 구성 요소와 캐릭터 컨테이너 구성 요소에서 저장형 교차 사이트 스크립팅XSS에 취약합니다. 공격자는 설명, 인사말, 예제 대화 또는 시스템 프롬프트LLM이 채팅 응답에 XSS 페이로드를 포함하도록 지시 중 하나에 SVG XSS 페이로드를 포함한 AI 캐릭터를 제작하여 임의의 클라이언트 측 스크립트 실행을 달성할 수 있습니다. 사용자가 이러한 악성 AI 캐릭터와 상호 작용하거나 해당 프로필을 탐색하기만 해도...

delirium-ai-safety-benchmark

Delirium AI 안전성 벤치마크 LLM의 ACE정서적 맥락 침식 및 관련 리미널 공격 벡터에 대한 취약성을 측정하는 진단 프레임워크. Delirium 은 악용 도구가 아니다. 언어 모델의 어텐션 가중치가 시스템 프롬프트를 따르는 상태에서 새롭게 출현한 대인 관계 패턴을 따르는 상태로 전환되는 정확한 순간을 — 피해가 발생하기 전에 — 감지하도록 설계된 표준화된 벤치마크이다. ⚠️ 윤리 헌장 1. 진단 전용 — Delirium은 탐지하고, 측정하고, 보고한다. 악용하지 않는다. 2. 킬 스위치 — 벤치마크가 3단계...

CVE-2026-82280

Quivr through 0.0.322 fails to validate ownership in its prompt endpoints, allowing any authenticated user to modify prompts by identifier. An attacker with only read-only access to a shared brain can discover exposed prompt identifiers and overwrite system prompts , affecting all users of that b...

EUVD-2026-67811

Quivr through 0.0.322 fails to validate ownership in prompt endpoints, allowing authenticated users to modify any prompt by identifier. Attackers with read-only access to shared brains can read exposed prompt identifiers and overwrite system prompts affecting all brain users...

EUVD-2026-60966

CodeWhale versions before 0.8.64 fail to validate file paths in the project config instructions field, allowing attackers to read arbitrary files on the victim's system. A malicious .codewhale/config.toml file in a cloned repository can specify paths outside the workspace that are read and inject...

CVE-2026-75859

CodeWhale before 0.8.64 fails to validate file paths in the project config instructions field. A malicious .codewhale/config.toml in a cloned repo can read files outside the workspace and inject their contents into the AI system prompt, enabling exfiltration. CVSS 7.5 HIGH. Patch released; no kno...

CVE-2026-75859 CodeWhale before 0.8.64 Arbitrary File Read via instructions

CodeWhale versions before 0.8.64 fail to validate file paths in the project config instructions field, allowing attackers to read arbitrary files on the victim's system. A malicious .codewhale/config.toml file in a cloned repository can specify paths outside the workspace that are read and inject...

CVE-2026-19011

CVE-2026-19011 affects TinyAGI 0.0.20. The vulnerable element is the function buildSystemPrompt in packages/server/src/routes/agents.ts, where manipulation can cause file inclusion. The attack may be initiated remotely and, per the description, the exploit is public. No remediation details are pr...

CVE-2026-19011 TinyAGI agents.ts buildSystemPrompt file inclusion

A vulnerability was detected in TinyAGI 0.0.20. The affected element is the function buildSystemPrompt of the file packages/server/src/routes/agents.ts. Performing a manipulation results in file inclusion. The attack may be initiated remotely. The exploit is now public and may be used. The projec...

EUVD-2026-53791

A vulnerability was detected in TinyAGI 0.0.20. The affected element is the function buildSystemPrompt of the file packages/server/src/routes/agents.ts. Performing a manipulation results in file inclusion. The attack may be initiated remotely. The exploit is now public and may be used. The projec...

CVE-2026-19011 TinyAGI agents.ts buildSystemPrompt file inclusion

A vulnerability was detected in TinyAGI 0.0.20. The affected element is the function buildSystemPrompt of the file packages/server/src/routes/agents.ts. Performing a manipulation results in file inclusion. The attack may be initiated remotely. The exploit is now public and may be used. The projec...

PYSEC-2026-2728 Open WebUI: Sharing models for others to use (read permission) also exposes model details (system prompt leakage)

Summary When setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may consider their system prompt confidential, so we consider this a security issue. Compare...

CVE-2026-15519

The CVE-2026-15519 entry affects usestrix strix up to 1.0.2, specifically the PyPI Handler’s file system_prompt.jinja. A manipulation can cause inclusion of functionality from an untrusted control sphere. The vulnerability is exploitable remotely with network impact and a high attack complexity; ...

CVE-2026-45387

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.5, when setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may...

CVE-2026-45387

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.5, when setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may...

CVE-2026-45351

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...

CVE-2026-45387

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.5, when setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may...

CVE-2026-45387 Open WebUI: Sharing models for others to use (read permission) also exposes model details (system prompt leakage)

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.5, when setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may...

EUVD-2026-30628

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.5, when setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may...