Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-112265 Overview
CVE-2026-112265 is a missing authenticity verification vulnerability [CWE-345] affecting LangBot through version 4.11.0. The LangBot Lark/Feishu adapter accepts inbound webhook events without validating the HMAC signature that Lark/Feishu platforms attach to each request. Unauthenticated attackers who learn a bot UUID can POST forged im.message.receive_v1 or card.action.trigger events directly to the webhook endpoint. These forged events drive the large language model (LLM) pipeline, tool invocations, and approval workflows while impersonating other users.
Critical Impact
Remote attackers can impersonate arbitrary users across LangBot's LLM pipeline, triggering unauthorized tool execution and approval actions without authentication.
Affected Products
- LangBot versions through 4.11.0
- LangBot Lark adapter (src/langbot/pkg/platform/adapters/lark/adapter.py)
- LangBot Lark source module (src/langbot/pkg/platform/sources/lark.py)
Discovery Timeline
- 2026-10-19 - CVE-2026-112265 published to NVD
- 2026-10-19 - Last updated in NVD database
Technical Details for CVE-2026-112265
Vulnerability Analysis
LangBot exposes a webhook controller that receives event callbacks from Lark/Feishu platforms. The affected endpoint processes incoming JSON payloads such as im.message.receive_v1 (incoming chat messages) and card.action.trigger (interactive card events). Lark/Feishu platforms sign each webhook request using an HMAC-based signature derived from a shared secret.
The LangBot adapter does not verify this signature before dispatching events into its message handling pipeline. Any party able to reach the webhook endpoint and guess or obtain a bot UUID can submit arbitrary payloads. The server treats these payloads as authentic events from Lark/Feishu.
This maps to [CWE-345]: Insufficient Verification of Data Authenticity. The impact spans forged user messages, invocation of LLM-connected tools, and manipulation of approval flows tied to specific user identities.
Root Cause
The root cause is the absence of HMAC signature verification in the webhook processing path. The affected modules did not import or use Python's hmac library when handling inbound Lark/Feishu events. Without cryptographic verification, request origin cannot be validated.
Attack Vector
An attacker requires knowledge of a target bot's UUID, which is passed in the webhook URL path. With the UUID known, the attacker crafts a JSON event mimicking the Lark/Feishu callback schema and POSTs it to the LangBot webhook endpoint. The forged event propagates through adapter logic, enters the LLM pipeline, and executes any downstream tools or approval actions as the spoofed user.
# Security patch for src/langbot/pkg/platform/adapters/lark/adapter.py
import asyncio
import base64
import hashlib
+import hmac
import json
import time
import traceback
Source: LangBot Commit 43c481d
# Security patch for src/langbot/pkg/platform/sources/lark.py
import traceback
import typing
import asyncio
+import hmac
import re
import base64
import uuid
Source: LangBot Commit 43c481d
Both patches introduce the hmac module, enabling constant-time signature comparison against the Lark/Feishu shared secret before events are accepted.
Detect Identity Attacks the Moment Authentication Fails
Singularity™ Identity surfaces credential misuse, lateral movement, and privilege escalation in real time, so you can contain the breach before it spreads.
Stop Identity AttacksDetection Methods for CVE-2026-112265
Indicators of Compromise
- Inbound POST requests to LangBot webhook paths (such as /webhook/lark/<bot_uuid>) from source IPs outside the official Lark/Feishu callback ranges.
- Webhook requests missing or carrying malformed X-Lark-Signature and X-Lark-Request-Timestamp headers.
- LLM tool invocations or approval actions attributed to users who did not send a corresponding chat message in the Lark/Feishu channel audit log.
Detection Strategies
- Compare LangBot webhook request logs against Lark/Feishu's documented callback IP ranges and alert on mismatches.
- Correlate im.message.receive_v1 events reaching the LLM pipeline with authenticated user activity in the upstream Lark/Feishu tenant.
- Flag any card.action.trigger event that lacks a prior rendered card session for the referenced user.
Monitoring Recommendations
- Enable verbose logging on the LangBot webhook controller (webhooks.py) to capture full request headers for forensic review.
- Monitor outbound actions from the LLM pipeline, including tool calls and approval decisions, for anomalous volume or unexpected user attribution.
- Alert on repeated webhook requests sharing the same bot UUID from distinct source IPs over short time windows.
How to Mitigate CVE-2026-112265
Immediate Actions Required
- Upgrade LangBot to the fixed version containing commit 43c481d74a6388657b84baa7d292c613c75c24f9, which adds HMAC signature verification.
- Restrict network access to the LangBot webhook endpoint using an allowlist of Lark/Feishu callback IP ranges at the firewall or reverse proxy.
- Rotate any bot UUIDs and Lark/Feishu application secrets suspected of exposure.
Patch Information
The maintainers merged pull request #2628 in commit 43c481d, titled fix/lark-webhook-signature-verification. The patch imports the hmac module in both src/langbot/pkg/platform/adapters/lark/adapter.py and src/langbot/pkg/platform/sources/lark.py and enforces signature validation on incoming webhook events. Review the LangBot Commit 43c481d, LangBot Issue #2627, and the VulnCheck Advisory for LangBot for the full remediation context.
Workarounds
- Place the LangBot webhook endpoint behind a reverse proxy that validates the Lark/Feishu HMAC signature before forwarding requests upstream.
- Treat bot UUIDs as sensitive credentials and avoid exposing them in client-side code, logs, or public configuration files.
- Disable LLM tool invocation and approval automation for Lark/Feishu channels until the patched release is deployed.
# Example nginx allowlist restricting webhook access to Lark/Feishu ranges
location /webhook/lark/ {
allow 203.0.113.0/24; # replace with official Lark/Feishu callback CIDRs
deny all;
proxy_pass http://langbot_upstream;
}
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

