CVE-2026-10224 Overview
CVE-2026-10224 is a resource consumption vulnerability [CWE-400] affecting NousResearch hermes-agent versions up to 2026.4.30. The flaw resides in the _handle_webhook_request function within gateway/platforms/feishu.py, part of the Webhook Endpoint component. Remote attackers can trigger the issue without authentication or user interaction, leading to availability degradation of the affected service. The exploit has been publicly disclosed, increasing the likelihood of opportunistic abuse. According to the disclosing party, the vendor was contacted prior to publication but did not respond.
Critical Impact
Unauthenticated remote attackers can submit crafted webhook requests to the Feishu platform handler, causing resource exhaustion and disrupting availability of the hermes-agent service.
Affected Products
- NousResearch hermes-agent up to and including version 2026.4.30
- Deployments exposing the Feishu webhook endpoint (gateway/platforms/feishu.py)
- Integrations relying on the _handle_webhook_request handler
Discovery Timeline
- 2026-06-01 - CVE-2026-10224 published to NVD
- 2026-06-01 - Last updated in NVD database
Technical Details for CVE-2026-10224
Vulnerability Analysis
The vulnerability resides in the _handle_webhook_request function of gateway/platforms/feishu.py in NousResearch hermes-agent. The handler processes inbound webhook requests from the Feishu messaging platform without sufficient controls on resource usage. Attackers can craft requests that force the agent to allocate or consume system resources disproportionate to the request size or complexity. Repeated or concurrent abuse drives CPU, memory, or connection consumption upward until the service becomes unresponsive. The issue is classified under [CWE-400] Uncontrolled Resource Consumption, and the EPSS score is 0.042% (percentile 13.045).
Root Cause
The handler does not enforce adequate validation, rate limiting, or bounds on incoming webhook payloads. Inputs accepted by _handle_webhook_request trigger processing paths that consume resources without proportional safeguards. Because the endpoint is reachable over the network and requires no authentication, the trust boundary at the webhook surface is too permissive for production exposure.
Attack Vector
An attacker delivers crafted HTTP requests to the exposed Feishu webhook endpoint. No credentials or user interaction are required. Repeated requests, or a single sufficiently large or malformed payload, cause the agent to exhaust available resources. Refer to the VulDB Vulnerability Details and the GitHub PoC Repository for technical specifics. No verified exploitation code is reproduced here.
Detection Methods for CVE-2026-10224
Indicators of Compromise
- Sudden CPU or memory spikes on hosts running hermes-agent without a corresponding increase in legitimate traffic
- Bursts of inbound POST requests to the Feishu webhook endpoint from one or a small set of source addresses
- Hermes-agent process restarts, hangs, or unhandled exception logs tied to webhook handling
- Increased latency or 5xx responses from the gateway service during webhook ingestion
Detection Strategies
- Baseline normal webhook request rates and payload sizes, then alert on statistical outliers
- Inspect HTTP request bodies reaching _handle_webhook_request for anomalous structure or size
- Correlate process-level resource telemetry with network ingress to identify amplification patterns
- Track repeated requests from the same client identifiers against the Feishu endpoint
Monitoring Recommendations
- Forward webhook gateway logs and host performance metrics to a centralized analytics pipeline
- Set thresholds on per-source request volume and concurrent connections to the Feishu handler
- Monitor file descriptor, thread, and memory usage of the hermes-agent process
- Alert on service restarts, OOM events, or watchdog timeouts on hosts running the agent
How to Mitigate CVE-2026-10224
Immediate Actions Required
- Restrict network exposure of the Feishu webhook endpoint to trusted source ranges only
- Place a reverse proxy or API gateway in front of hermes-agent to enforce request size limits and rate limiting
- Validate Feishu webhook signatures and reject unsigned or malformed requests before they reach _handle_webhook_request
- Monitor hermes-agent hosts for resource exhaustion and prepare automated restart or failover procedures
Patch Information
At the time of publication, no vendor patch has been referenced in the available advisories. The disclosing party reports that the vendor did not respond to contact attempts. Operators should track the VulDB CVE Report and the upstream NousResearch hermes-agent repository for an official fix and apply it once released.
Workarounds
- Disable the Feishu webhook integration if it is not required in the deployment
- Terminate TLS and apply Web Application Firewall (WAF) rules that cap request rate, body size, and concurrency per source
- Run hermes-agent under strict cgroup or container resource limits to contain the impact of resource exhaustion
- Require mutual TLS or a shared secret in front of the webhook endpoint to block unauthenticated traffic
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
