Vulnerability  ·  2026-10-20

LangBot Lark/Feishu webhook authenticity bypass → drive LLM pipeline and approvals as other users

VulnerabilityMedium impactGlobalCVE-2026-112265
VulnCheck published (2026-10-19) that LangBot through 4.11.0 fails to verify the Lark/Feishu webhook signature, allowing unauthenticated attackers to inject forged im.message.receive_v1 / card.action.trigger events. CVSS 6.5 (v3.1) / 8.3 (v4.0), CWE-345.
Chat-connector webhooks are a common entry point into agent workflows. Forging a legitimate-looking chat event lets an outsider drive an enterprise LangBot agent to run tools, trigger approvals, and answer as other users, bypassing the chat channel's authenticity boundary.
The Lark/Feishu webhook controller skips signature verification (CWE-345 missing authenticity verification). An attacker knowing a bot UUID POSTs fabricated events to the webhook endpoint, injecting messages into the LLM conversation pipeline and triggering tools and approval flows on behalf of legitimate users.
LangBot through 4.11.0
Update LangBot to a version incorporating the fix commit (43c481d74a6388657b84baa7d292c613c75c24f9).
NVD detailLangBot webhook controllerVulnCheck advisory
See this in the live feed Explore related AI security and governance findings — updated every morning.
Open the feed →