OpenHuman: MCP tool description injection: untrusted server descriptions execute in agent LLM context without sanitization
Affects OpenHuman versions before 0.57.13 in setups with any MCP server configured, including community servers. CVSS scores it 8.3 (high). Fixed in 0.57.13.
Written by anthropic/claude-haiku-5.5 from the sources below. Verify against them.
- Exploited
- No known exploitation
- Fix
- Fixed in 0.57.13
- Who is exposed
- Specific setups only
Do first: Check whether your setup matches what the bug needs. Victim has any MCP server configured in OpenHuman, including community servers.
- Published
- Last modified
- A 0-100 score for how much a CVE matters across the industry: exploitation, exposure, how widely the software is used, severity and whether a fix exists.What do these mean? →
- 20/100 · provisional
- How many real-world setups can hit the bug, from the vendor's advisory and CISA data. It does not know your setup.What do these mean? →
- Specific setups only
- Whether working attack code is public (Exploit-DB, Metasploit, Nuclei or GitHub). Public code makes attacks easier.What do these mean? →
- None on Exploit-DB, Metasploit or GitHub
Summary
OpenHuman, an open-source agent harness sold as a desktop application and Rust library, is affected by a flaw where descriptions and other text from configured external MCP servers are passed into the agent's model context without sanitization. An attacker running or compromising such a server can remotely, without logging in, but only if a user chooses to interact with the malicious content, plant instructions that can steer the agent into invoking other tools, disclosing accessible data, or taking unauthorized actions.
Read the original description
OpenHuman is an open-source agent harness available as a desktop application and Rust library. Before 0.57.13, OpenHuman places names, descriptions, titles, and parameter definitions returned by configured external MCP servers into the agent model context without neutralizing instructions or bounding the descriptive fields. The McpRemoteTool records produced by list_tools preserve this remote content, while tool filtering considers the tool name and the prompt-injection detector screens user messages rather than MCP definitions. A malicious or compromised MCP server can therefore return a poisoned description that is evaluated on every agent turn and may direct the model to invoke other available tools, disclose accessible data, or perform unauthorized actions without a separate malicious user message. The issue affects users who configure an attacker-controlled or compromised MCP server. This issue is fixed in version 0.57.13.
cve.now Index
- Exploitation?/40Unknown. No CISA SSVC exploitation rating yet
- Exposure12/25Specific setups only, reachable over the network without a login
- Reach?/20Unknown. No tracked technology or package matched
- Severity8.3/10CVSS 8.3
- No patch0/5No signal
What would change the score
- Would score about 70 if added to CISA KEV
- Would score about 65 if CISA's SSVC rates exploitation active
What to do
Check whether your setup matches what the bug needs.
Victim has any MCP server configured in OpenHuman, including community servers.Upgrade to a fixed version: 0.57.13.
Timeline
- CVE published
- First seen by cve.now
Affected versions
- tinyhumansai openhuman
fixed in 0.57.13
Evidence 4
- Advisory
- Repository
- Repository
- Repository
Corrections
None so far.
Something wrong on this page? Report an error. Every fix is listed here.
Questions about CVE-2026-54140
Is CVE-2026-54140 being exploited?
It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.Source: cve.now feed data
What does this mean?
A malicious or compromised MCP server can place arbitrary LLM instructions in a tool description. OpenHuman passes these descriptions into the agent's model context without sanitization, so the embedded instructions run on every agent turn.Source: github.com
Which versions are affected by CVE-2026-54140?
openhuman: before 0.57.13.Source: cve.now feed data
Am I affected, and how do I check?
OpenHuman builds before 0.57.13 are affected in setups where any MCP server is configured, including community servers. To check, compare the installed OpenHuman version with 0.57.13 and review whether any MCP servers are configured.Source: github.com
How do I fix CVE-2026-54140?
Upgrade OpenHuman to 0.57.13 or later, which is the listed fixed version.Source: github.com