<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai-Gateway — PoC Archive</title><link>https://poc.intelseclab.com/tags/ai-gateway/</link><description>Latest proof-of-concept entries.</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sun, 12 Jul 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://poc.intelseclab.com/tags/ai-gateway/index.xml" rel="self" type="application/rss+xml"/><item><title>Flowise Enterprise Authentication Bypass via Hardcoded Default JWT Secrets (CVE-2026-56271)</title><link>https://poc.intelseclab.com/pocs/web/2026-07-12_cve-2026-56271-flowise-hardcoded-jwt-authbypass/</link><pubDate>Sun, 12 Jul 2026 00:00:00 +0000</pubDate><guid>https://poc.intelseclab.com/pocs/web/2026-07-12_cve-2026-56271-flowise-hardcoded-jwt-authbypass/</guid><description>Critical severity (CVSS 9.8) — web · CVE-2026-56271 (GHSA-cc4f-hjpj-g9p8). Status: Weaponized (functional PoC forges valid admin JWTs and confirms bypass against real endpoints). Affects: Flowise — open-source low-code LLM/agent orchestration platform (enterprise edition, passport authentication middleware). Tags: flowise, ai-gateway, llm-orchestration, jwt, hardcoded-secret, authentication-bypass, cwe-321, unauthenticated, remote, privilege-escalation.</description><category>web</category><category>Critical</category><category>flowise</category><category>ai-gateway</category><category>llm-orchestration</category><category>jwt</category><category>hardcoded-secret</category><category>authentication-bypass</category><category>cwe-321</category><category>unauthenticated</category><category>remote</category><category>privilege-escalation</category></item><item><title>LiteLLM Proxy Pre-Authentication SQL Injection via Error-Handling Callback (CVE-2026-42208)</title><link>https://poc.intelseclab.com/pocs/web/2026-07-11_cve-2026-42208-litellm-sqli-proxy-authbypass/</link><pubDate>Sat, 11 Jul 2026 00:00:00 +0000</pubDate><guid>https://poc.intelseclab.com/pocs/web/2026-07-11_cve-2026-42208-litellm-sqli-proxy-authbypass/</guid><description>Critical severity (CVSS 9.8) — web · CVE-2026-42208 (GHSA-r75f-5x8p-qvmc). Status: Weaponized (public working PoC + Docker lab, actively exploited in the wild within 36 hours of disclosure). Affects: LiteLLM Proxy — open-source LLM/AI gateway (22,000+ GitHub stars) fronting OpenAI, Anthropic, and other model provider APIs. Tags: litellm, ai-gateway, llm-proxy, sql-injection, cwe-89, unauthenticated, remote, blind-sqli, kev-adjacent, credential-theft.</description><category>web</category><category>Critical</category><category>litellm</category><category>ai-gateway</category><category>llm-proxy</category><category>sql-injection</category><category>cwe-89</category><category>unauthenticated</category><category>remote</category><category>blind-sqli</category><category>kev-adjacent</category><category>credential-theft</category></item><item><title>LiteLLM Authentication Bypass via OIDC Userinfo Cache Key Collision (CVE-2026-35030)</title><link>https://poc.intelseclab.com/pocs/web/2026-07-05_cve-2026-35030-litellm-oidc-cache-collision-authbypass/</link><pubDate>Sun, 05 Jul 2026 00:00:00 +0000</pubDate><guid>https://poc.intelseclab.com/pocs/web/2026-07-05_cve-2026-35030-litellm-oidc-cache-collision-authbypass/</guid><description>Critical severity (CVSS 9.1) — web · CVE-2026-35030. Status: Weaponized. Affects: LiteLLM proxy (with enable_jwt_auth: true). Tags: authentication-bypass, litellm, oidc, jwt, cache-collision, ai-gateway, cwe-287.</description><category>web</category><category>Critical</category><category>authentication-bypass</category><category>litellm</category><category>oidc</category><category>jwt</category><category>cache-collision</category><category>ai-gateway</category><category>cwe-287</category></item><item><title>LiteLLM /config/update Broken Access Control (CVE-2026-35029)</title><link>https://poc.intelseclab.com/pocs/web/2026-07-05_cve-2026-35029-litellm-config-broken-access-control/</link><pubDate>Sun, 05 Jul 2026 00:00:00 +0000</pubDate><guid>https://poc.intelseclab.com/pocs/web/2026-07-05_cve-2026-35029-litellm-config-broken-access-control/</guid><description>High severity (CVSS 8.8) — web · CVE-2026-35029. Status: Weaponized. Affects: LiteLLM proxy. Tags: broken-access-control, litellm, ai-gateway, config-update, ssrf, rce, environment-variable-theft, cwe-863.</description><category>web</category><category>High</category><category>broken-access-control</category><category>litellm</category><category>ai-gateway</category><category>config-update</category><category>ssrf</category><category>rce</category><category>environment-variable-theft</category><category>cwe-863</category></item><item><title>Authenticated Command Injection in LiteLLM MCP Test Endpoints (CVE-2026-42271)</title><link>https://poc.intelseclab.com/pocs/web/2026-07-01_cve-2026-42271-litellm-mcp-command-injection/</link><pubDate>Wed, 01 Jul 2026 00:00:00 +0000</pubDate><guid>https://poc.intelseclab.com/pocs/web/2026-07-01_cve-2026-42271-litellm-mcp-command-injection/</guid><description>High severity (CVSS 8.7) — web · CVE-2026-42271. Status: PoC. Affects: BerriAI LiteLLM (proxy) — MCP preview/test endpoints. Tags: command-injection, LiteLLM, MCP, AI-application, ai-gateway, authenticated, api-key, subprocess.</description><category>web</category><category>High</category><category>command-injection</category><category>LiteLLM</category><category>MCP</category><category>AI-application</category><category>ai-gateway</category><category>authenticated</category><category>api-key</category><category>subprocess</category></item></channel></rss>