EUVD-2026-58032
GHSA-42r3-x6vx-x49x
Severity by source
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N
Pure availability DoS via network-reachable input; no confidentiality or integrity impact - C:H in provided vector is a scoring error.
Primary rating from Vendor (mitre).
CVSS VectorVendor: mitre
Lifecycle Timeline
4DescriptionCVE.org
crmne/ruby_llm at commit fa6f279847d6d7027814539d9c0dfc3bbdfd2a83 contains a polynomial-time regular expression denial-of-service condition in RubyLLM::Utils.underscore on Ruby 3.1.x. A very long crafted class, agent, or tool name can cause excessive CPU consumption and a denial of service.
AnalysisAI
Polynomial regular expression denial-of-service in the crmne/ruby_llm Ruby gem allows remote unauthenticated attackers to exhaust CPU resources by submitting a crafted, excessively long class, agent, or tool name that triggers the inefficient regex in RubyLLM::Utils.underscore. The vulnerability is constrained to Ruby 3.1.x environments due to engine-specific regex behavior at that commit. A public proof-of-concept exists (hosted as a GitHub Gist), though no active exploitation has been confirmed by CISA KEV.
Technical ContextAI
RubyLLM is a Ruby gem providing a client abstraction layer for LLM APIs (OpenAI, Anthropic, etc.). The vulnerable code resides in lib/ruby_llm/utils.rb at line 13 of commit fa6f279847d6d7027814539d9c0dfc3bbdfd2a83, specifically in the Utils.underscore method, which uses a regular expression to convert CamelCase strings to snake_case. On Ruby 3.1.x, the regex engine's backtracking behavior when applied to pathologically long input strings results in polynomial time complexity rather than linear time - this is classified under CWE-1333 (Inefficient Regular Expression Complexity). The affected CPE is listed as cpe:2.3:a:n/a:n/a:*:*:*:*:*:*:*:* (non-standardized), indicating the affected product has not been formally catalogued in NVD's product dictionary. The attack surface depends on whether the application passes user-controlled strings as class, agent, or tool names into the RubyLLM pipeline.
RemediationAI
Upgrade to a commit or release of ruby_llm that patches the Utils.underscore regex - the exact fixed release version is not independently confirmed from available data; consumers should check the upstream repository at https://github.com/crmne/ruby_llm for a tagged release addressing this issue. As a compensating control until patching is possible, applications should enforce strict length limits on any user-supplied class, agent, or tool names before they reach RubyLLM::Utils.underscore - for example, reject inputs exceeding a few hundred characters at the API boundary. Additionally, consider upgrading the Ruby runtime beyond 3.1.x, as the regex behavior responsible for polynomial backtracking may differ in later Ruby versions. Rate-limiting API endpoints that accept name inputs limits the blast radius of repeated DoS attempts. Note that input length validation has no functional side effects for legitimate use cases, since valid class and agent names are rarely longer than a few dozen characters.
LibreNMS before 24.10.0 allows a remote attacker to execute arbitrary code via OS command injection involving AboutContr
Reachable assertion in Open5GS up to version 2.7.6 allows remote unauthenticated attackers to crash the Diameter CER han
A Command Injection issue in the payload build page in BYOB (Build Your Own Botnet) 2.0 allows attackers to execute arbi
Denial of service in Open5GS 2.7.6 and earlier allows network-adjacent attackers with low-privilege Diameter Gx peer acc
Denial of service in Open5GS HSS (Home Subscriber Server) versions 2.7.0 through 2.7.6 is triggered remotely by manipula
Remote command injection in Control Web Panel allows unauthenticated attackers to execute arbitrary OS commands as root
LINQPad before 5.52.01 Pro edition is vulnerable to Unsafe Deserialization in LINQPad.AutoRefManager::PopulateFromCache(
Remote code execution in dash-uploader (Python package for Plotly Dash) versions 0.1.0 through 0.7.0a2 allows unauthenti
Command injection in LiteLLM 1.18.10 lets attackers who can configure MCP (Model Context Protocol) servers supply arbitr
SQL injection in AzuraCast's backup restore flow (versions up to and including 0.23.7) lets an attacker achieve privileg
Arbitrary file read in Yu AI Code Mother v4.3 lets anonymous, remote attackers retrieve files outside the intended previ
Remote code execution in APScheduler (all versions through 3.10.x and 4.0.0a5) is achievable when applications deseriali
Same technique Denial Of Service
View allShare
External POC / Exploit Code
Leaving vuln.today