Blinko is an AI-powered card note-taking project. Prior to version 1.8.4, the MCP (Model Context Protocol) server creation function allows specifying arbitrary commands and arguments, which are executed when testing the connection. This issue has been patched in version 1.8.4.
mcp-memory-service is an open-source memory backend for multi-agent systems. Prior to version 10.25.1, when the HTTP server is enabled (MCP_HTTP_ENABLED=true), the application configures FastAPI's CORSMiddleware with allow_origins=['*'], allow_credentials=True, allow_methods=["*"], and allow_headers=["*"]. The wildcard Access-Control-Allow-Origin: * header permits any website to read API responses cross-origin. When combined with anonymous access (MCP_ALLOW_ANONYMOUS_ACCESS=true) - the simplest way to get the HTTP dashboard working without OAuth - no credentials are needed, so any malicious website can silently read, modify, and delete all stored memories. This issue has been patched in version 10.25.1.
A vulnerability was found in sigmade Git-MCP-Server up to 785aa159f262a02d5791a5d8a8e13c507ac42880. Affected by this vulnerability is the function child_process.exec of the file src/gitUtils.ts of the component show_merge_diff/quick_merge_summary/show_file_diff. The manipulation results in os command injection. The attack must be initiated from a local position. The exploit has been made public and could be used. This product operates on a rolling release basis, ensuring continuous delivery. Consequently, there are no version details for either affected or updated releases. It is advisable to implement a patch to correct this issue. The vendor was contacted early about this disclosure but did not respond in any way.
CKAN MCP Server is a tool for querying CKAN open data portals. Versions prior to 0.4.85 provide tools including ckan_package_search and sparql_query that accept a base_url parameter, making HTTP requests to arbitrary endpoints without restriction. A CKAN portal client has no legitimate reason to contact cloud metadata or internal network services. There is no URL validation on base_url parameter. No private IP blocking (RFC 1918, link-local 169.254.x.x), no cloud metadata blocking. The sparql_query and ckan_datastore_search_sql tools also accept arbitrary base URLs and expose injection surfaces. An attack can lead to internal network scanning, cloud metadata theft (IAM credentials via IMDS at 169.254.169.254), potential SQL/SPARQL injection via unsanitized query parameters. Attack requires prompt injection to control the base_url parameter. This issue has been fixed in version 0.4.85.
CVE-2026-4270 is a medium severity vulnerability in AWS API MCP Server versions from 0.2.14 up to but not including 1.3.9. It involves improper protection of an alternate path in the no-access and workdir feature, which may allow bypassing intended file access restrictions and expose arbitrary local file contents within the MCP client application context. The issue is fixed in version 1.3.9.
CVE-2025-69902 is a critical command injection vulnerability found in the minimal_wrapper.py component of kubectl-mcp-server version 1.2.0. This flaw allows unauthenticated remote attackers to execute arbitrary shell commands by injecting malicious shell metacharacters. The vulnerability has a CVSS score of 9.8, reflecting its high impact on confidentiality, integrity, and availability without requiring user interaction or privileges. Although no known exploits are currently reported in the wild, the ease of exploitation and the critical nature of the vulnerability make it a significant threat. Organizations using kubectl-mcp-server should prioritize patching or mitigating this issue to prevent potential compromise. The vulnerability is related to CWE-94, indicating improper control of code injection.
CVE-2026-4198 is a command injection vulnerability in hypermodel-labs mcp-server-auto-commit version 1.0.0, specifically in the getGitChanges function of index.ts. The flaw allows an attacker with local access and low privileges to execute arbitrary commands on the host system. Exploitation does not require user interaction but is limited to local attack vectors. Although the vulnerability has a medium CVSS score of 4.8, it poses a risk to confidentiality, integrity, and availability due to potential command execution. The vendor has been notified but has not yet responded, and a patch is available to remediate the issue. No known exploits are currently in the wild.
CVE-2026-4192 is a command injection vulnerability found in version 1.0.0 of AvinashBole's quip-mcp-server, specifically in the setupToolHandlers function within src/index.ts. This flaw allows remote attackers to execute arbitrary commands on the affected server without requiring user interaction or elevated privileges beyond low-level privileges. The vulnerability has been publicly disclosed, but no patch or vendor response is currently available. The CVSS 4.0 base score is 5.3, indicating a medium severity level. Exploitation could lead to partial compromise of confidentiality, integrity, and availability of the server.
CVE-2026-3959 is an OS command injection vulnerability in the 0xKoda WireMCP product, specifically in the Tshark CLI Command Handler's server.tool function within index.js. The flaw allows an attacker with local access and low privileges to execute arbitrary OS commands by manipulating inputs to this function. Exploitation does not require user interaction but does require local access and low complexity. The vulnerability has a medium severity with a CVSS score of 4.8. No patch or vendor response has been provided yet, and the product uses a rolling release system complicating version tracking. While no known exploits are currently in the wild, the exploit details have been publicly disclosed. Organizations using WireMCP should be cautious of local threat actors leveraging this vulnerability to escalate privileges or execute malicious commands.
CVE-2026-32112 is a cross-site scripting (XSS) vulnerability in the ha-mcp OAuth consent form of the Home Assistant MCP Server prior to version 7.0.0. The vulnerability arises because user-controlled parameters are rendered using Python f-strings without proper HTML escaping, allowing injection of malicious JavaScript. Exploitation requires an attacker to reach the OAuth endpoint and trick the server operator into visiting a crafted authorization URL. This issue affects only users who have explicitly enabled the beta OAuth mode (ha-mcp-oauth), which is not part of the default setup. The vulnerability has a CVSS score of 6.8, indicating medium severity, with high impact on confidentiality and integrity but no impact on availability. No known exploits are currently in the wild. The vulnerability is fixed in version 7.
CVE-2026-32111 is a Server-Side Request Forgery (SSRF) vulnerability in the ha-mcp component of homeassistant-ai versions prior to 7.0.0. The vulnerability arises because the OAuth consent form accepts a user-supplied ha_url and makes an HTTP request to {ha_url}/api/config without validating the URL. This allows unauthenticated attackers to perform internal network reconnaissance by submitting arbitrary URLs. Two other OAuth-related code paths (REST and WebSocket) are also affected by this SSRF primitive. The primary deployment method using a private URL with a pre-configured HOMEASSISTANT_TOKEN is not vulnerable. The issue has a CVSS score of 5.3 (medium severity) and was fixed in version 7.0.
MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to version 0.17.0, the `confluence_download_attachment` MCP tool accepts a `download_path` parameter that is written to without any directory boundary enforcement. An attacker who can call this tool and supply or access a Confluence attachment with malicious content can write arbitrary content to any path the server process has write access to. Because the attacker controls both the write destination and the written content (via an uploaded Confluence attachment), this constitutes for arbitrary code execution (for example, writing a valid cron entry to `/etc/cron.d/` achieves code execution within one scheduler cycle with no server restart required). Version 0.17.0 fixes the issue.
CVE-2026-27826 is a high-severity Server-Side Request Forgery (SSRF) vulnerability in sooperset's mcp-atlassian server versions prior to 0.17.0. It allows unauthenticated attackers who can access the HTTP endpoint to force the server to make arbitrary outbound HTTP requests by supplying two custom headers without needing an Authorization header. This vulnerability resides in the HTTP middleware and dependency injection layer, bypassing tool-level code detection. In cloud environments, it can lead to theft of IAM role credentials via the instance metadata service. In any deployment, it enables internal network reconnaissance and injection of attacker-controlled content into LLM tool results. The issue is fixed in version 0.17.0.
A security flaw has been discovered in RyuzakiShinji biome-mcp-server up to 1.0.0. Affected by this issue is some unknown functionality of the file biome-mcp-server.ts. Performing a manipulation results in command injection. The attack can be initiated remotely. The exploit has been released to the public and may be used for attacks. The patch is named 335e1727147efeef011f1ff8b05dd751d8a660be. Applying a patch is the recommended action to fix this issue.
WeKnora is an LLM-powered framework designed for deep document understanding and semantic retrieval. Prior to version 0.3.0, a vulnerability involving tool name collision and indirect prompt injection allows a malicious remote MCP server to hijack tool execution. By exploiting an ambiguous naming convention in the MCP client (mcp_{service}_{tool}), an attacker can register a malicious tool that overwrites a legitimate one (e.g., tavily_extract). This enables the attacker to redirect LLM execution flow, exfiltrate system prompts, context, and potentially execute other tools with the user's privileges. This issue has been patched in version 0.3.0.
The shell tool within GitHub Copilot CLI versions prior to and including 0.0.422 can allow arbitrary code execution through crafted bash parameter expansion patterns. An attacker who can influence the commands executed by the agent (e.g., via prompt injection through repository files, MCP server responses, or user instructions) can exploit bash parameter transformation operators to execute hidden commands, bypassing the safety assessment that classifies commands as "read-only." This has been patched in version 0.0.423.
The vulnerability stems from how the CLI's shell safety assessment evaluates commands before execution. The safety layer parses and classifies shell commands as either read-only (safe) or write-capable (requires user approval). However, several bash parameter expansion features can embed executable code within arguments to otherwise read-only commands, causing them to appear safe while actually performing arbitrary operations.
The specific dangerous patterns are ${var@P}, ${var=value} / ${var:=value}, ${!var}, and nested $(cmd) or <(cmd) inside ${...} expansions. An attacker who can influence command text sent to the shell tool - for example, through prompt injection via malicious repository content (README files, code comments, issue bodies), compromised or malicious MCP server responses, or crafted user instructions containing obfuscated commands - could achieve arbitrary code execution on the user's workstation. This is possible even in permission modes that require user approval for write operations, since the commands can appear to use only read-only utilities to ultimately trigger write operations. Successful exploitation could lead to data exfiltration, file modification, or further system compromise.
A vulnerability was detected in PhialsBasement nmap-mcp-server up to bee6d23547d57ae02460022f7c78ac0893092e38. Affected by this issue is the function child_process.exec of the file src/index.ts of the component Nmap CLI Command Handler. The manipulation results in command injection. The attack may be performed from remote. This product utilizes a rolling release system for continuous delivery, and as such, version information for affected or updated releases is not disclosed. The patch is identified as 30a6b9e1c7fa6146f51e28d6ab83a2568d9a3488. It is best practice to apply a patch to resolve this issue.
Golioth Pouch version 0.1.0, prior to commit 1b2219a1, contains a heap-based buffer overflow in BLE GATT server certificate handling. server_cert_write() allocates a heap buffer of size CONFIG_POUCH_SERVER_CERT_MAX_LEN when receiving the first fragment, then appends subsequent fragments using memcpy() without verifying that sufficient capacity remains. An adjacent BLE client can send unauthenticated fragments whose combined size exceeds the allocated buffer, causing a heap overflow and crash; integrity impact is also possible due to memory corruption.
Golioth Firmware SDK version 0.10.0 prior to 0.22.0, fixed in commit 48f521b, contain a stack-based buffer overflow in Payload Utils. The golioth_payload_as_int() and golioth_payload_as_float() helpers copy network-supplied payload data into fixed-size stack buffers using memcpy() with a length derived from payload_size. The only length checks are guarded by assert(); in release builds, the asserts are compiled out and memcpy() may copy an unbounded payload_size. Payloads larger than 12 bytes (int) or 32 bytes (float) can overflow the stack, resulting in a crash/denial of service. This is reachable via LightDB State on_payload with a malicious server or MITM.
Model Context Protocol Servers is a collection of reference implementations for the model context protocol (MCP). In mcp-server-git versions prior to 2026.1.14, the git_add tool did not validate that file paths provided in the files argument were within the repository boundaries. Because the tool used GitPython's repo.index.add() rather than the Git CLI, relative paths containing `../` sequences that resolve outside the repository were accepted and staged into the Git index. Users are advised to upgrade to 2026.1.14 or newer to remediate this issue.