VulnWatch VulnWatch
← Back to dashboard
#

Agentic / MCP

653 entries

Every Agentic / MCP entry VulnWatch has indexed, sorted by publication date.

Subscribe to this tag's RSS feed

High github

mcp-shell has a Secure Mode Allowlist Bypass via Default `/bin/bash` Executable

### Summary `mcp-shell` ships a default Docker configuration (`security.yaml`) that includes `/bin/bash` in the `allowed_executables` allowlist. The command validator (`security.go`) only checks whet...

8.4
CVSS
4 weeks ago
High github

mcp-shell has a Secure Mode Allowlist Bypass via Git Shell Alias

### Summary `mcp-shell`'s "secure mode" is designed to restrict command execution to an allowlist of executables defined in `security.yaml`. The default configuration includes `/usr/bin/git`. The sec...

8.4
CVSS
4 weeks ago
Critical github

qwed-mcp has Unsafe SymPy `parse_expr()` Remote Code Execution via Unsanitized Math Expression Input

### Summary `verify_math_expression()` in `qwed-mcp` v0.2.0 passes attacker-controlled strings directly to SymPy's `parse_expr()` without restricting `global_dict` or validating the expression's AST....

9.8
CVSS
4 weeks ago
High github

PraisonAI: Origin-validation bypass (startswith prefix match) enables unauthenticated cross-site request forgery against the PraisonAI MCP HTTP server

### Summary The PraisonAI MCP server exposes an HTTP-stream transport (praisonai mcp serve --transport http-stream) that binds to localhost and, by default, has no API key. Its only access control fo...

7.6
CVSS
4 weeks ago
Medium nvd

CVE-2026-55529: PraisonAI is a multi-agent teams system. Prior to praisonai 4.6.58, the MCP HTTP Stream _validate_origin method accepts

PraisonAI is a multi-agent teams system. Prior to praisonai 4.6.58, the MCP HTTP Stream _validate_origin method accepts request_origin.startswith(allowed), so the attacker-controlled localhost.evil.ex...

Agentic / MCP mcp server
6.9
CVSS
4 weeks ago
Medium github

praisonaiagents: ast_grep_rewrite rewrites arbitrary files without the @require_approval gate enforced on every sibling mutation tool

**Target:** PraisonAI (`MervinPraison/PraisonAI`) **Affected component:** `praisonaiagents/tools/ast_grep_tool.py` — `ast_grep_rewrite` **Affected versions:** master at `ce97667156a116c50b4a3d1aa21e09...

Agentic / MCP llm agent llm
6.1
CVSS
4 weeks ago
Medium github

PraisonAI MCP HTTP server has unauthenticated unbounded session accumulation (memory exhaustion; session TTL never enforced)

### Summary The PraisonAI MCP HTTP-stream server creates a new in-memory session on every initialize request and never removes it. The cleanup routine that would expire sessions (_cleanup_sessions) i...

Agentic / MCP mcp server
6.5
CVSS
4 weeks ago
Medium github

PraisonAI has an origin validation bypass in MCP HTTP Stream transport that allows browser-mediated unauthenticated tool execution on local MCP server

### Summary PraisonAI's MCP HTTP Stream transport uses an unsafe prefix match when validating the `Origin` header. The default localhost allowlist includes origins such as `http://localhost`, and the...

Agentic / MCP mcp server litellm
6.9
CVSS
4 weeks ago
Unknown rss_thehackernews

A Malicious Webpage Could Poison Your Local AI Model Behind NVIDIA NemoClaw

Oasis Security has disclosed a weakness in NVIDIA NemoClaw that could let an attacker-controlled webpage take unauthenticated control of the local Ollama instance serving an AI agent and plant hidden...

Agentic / MCP ai agent ai model ollama
4 weeks ago
Unknown rss_thehackernews

Marimo Notebook Flaw Could Run MCP Commands Before Cells Execute in Edit Mode

Marimo has addressed a high-severity security flaw in its notebook software that allowed an attacker to execute an attacker-supplied Model Context Protocol (MCP) command in a specially crafted noteboo...

Agentic / MCP model context protocol
4 weeks ago
Critical github

mcp-contextforge-gateway has RestrictedPython sandbox bypass via getattr builtin in python_sandbox_server

**Commit:** `f855e54d5b7bc1c91b977574a03b91eff6b86bb6` **Component:** `mcp-servers/python/python_sandbox_server/src/python_sandbox_server/server_fastmcp.py` ## Vulnerability RestrictedPython's sandb...

Agentic / MCP mcp server
10.0
CVSS
4 weeks ago
Medium nvd

CVE-2026-53509: CKAN MCP Server is a tool for querying CKAN open data portals. A known vulnerability CVE-2026-33060 indicated tools incl

CKAN MCP Server is a tool for querying CKAN open data portals. A known vulnerability CVE-2026-33060 indicated tools including ckan_package_search and sparql_query that accept a base_url parameter had...

5.7
CVSS
1 month ago
High nvd

CVE-2026-62677: Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, an authe

Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, an authenticated user can upload a session-scoped agent bundle with an absolute or trave...

8.8
CVSS
1 month ago
High nvd

CVE-2026-62676: Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, the shar

Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, the shared shell-command parser in omnigent/policies/builtins/_shell.py fails to recogni...

7.1
CVSS
1 month ago
High nvd

CVE-2026-62675: Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, multipar

Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, multipart POST /v1/sessions accepts an authenticated user's agent bundle and omnigent/se...

8.8
CVSS
1 month ago
Critical nvd

CVE-2026-62674: Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, PUT /ses

Omnigent is an open-source AI agent framework and meta-harness for orchestrating coding agents. Prior to 0.3.0, PUT /sessions/{session_id}/agent checks LEVEL_EDIT permission for a session but does not...

Agentic / MCP mcp server ai agent
9.0
CVSS
1 month ago
Medium nvd

CVE-2026-72846: Lightdash stores the webhook URL supplied with a scheduled delivery and later posts to it from sendWebhook in packages/b

Lightdash stores the webhook URL supplied with a scheduled delivery and later posts to it from sendWebhook in packages/backend/src/clients/GoogleChat/GoogleChatClient.ts and in packages/backend/src/cl...

5.3
CVSS
1 month ago
High github

Laravel Backpack CRUD: OS command injection in Stats::makeCurlRequest via attacker-controlled Host header (pre-auth)

## Summary `Backpack\CRUD\Stats::makeCurlRequest` builds a shell command using unescaped input that originates from the HTTP `Host` header, then passes it to `exec()`. A specially crafted Host header...

8.1
CVSS
1 month ago
High github

Laravel Backpack CRUD: Unverified password change in MyAccountController via mass assignment

## Summary The `MyAccountController::postAccountInfoForm` action bound to `POST /admin/edit-account-info` calls `$this->guard()->user()->update($request->except(['_token']))`. Because the controller...

7.6
CVSS
1 month ago
High nvd

CVE-2026-54449: LangBot is a global IM bot platform designed for LLMs. In version 4.10.7 and earlier, any authenticated user can add or

LangBot is a global IM bot platform designed for LLMs. In version 4.10.7 and earlier, any authenticated user can add or change an STDIO MCP server configuration without an adequate authorization bound...

Agentic / MCP mcp server llm
8.8
CVSS
1 month ago
Unknown nvd

CVE-2026-18482: Neo.mjs contains a command injection vulnerability within the FileSystemService.mjs component of the ai/mcp/server/file-

Neo.mjs contains a command injection vulnerability within the FileSystemService.mjs component of the ai/mcp/server/file-system MCP server, where the checkSyntax() and runPlaywrightTest() functions uns...

1 month ago
Unknown rss_thehackernews

Why "Shady AI" is Security's Next Big Governance Problem

In March 2026, an internal AI agent at Meta triggered a “Sev 1” incident after sensitive company and user data was exposed to employees who weren’t authorized to access it.  The incident began when a...

1 month ago
Medium nvd

CVE-2026-17153: The AI Agent by SiteGround plugin for WordPress is vulnerable to authorization bypass in all versions up to, and includi

The AI Agent by SiteGround plugin for WordPress is vulnerable to authorization bypass in all versions up to, and including, 1.2.7. This is due to the plugin not properly verifying that a user is autho...

5.3
CVSS
1 month ago
Critical nvd

CVE-2026-76404: In Splunk MCP Server app versions below 1.2.1, a user who holds the "admin" Splunk role could execute arbitrary commands

In Splunk MCP Server app versions below 1.2.1, a user who holds the "admin" Splunk role could execute arbitrary commands on the underlying operating system. The vulnerability is possible because of mi...

Agentic / MCP mcp server
9.1
CVSS
1 month ago
Medium github

SearXNG MCP Server: Additional hardened-mode SSRF bypasses

## Summary `mcp-searxng` has a hardened-mode URL-reading feature intended to prevent `web_url_read` from reaching private or internal network resources. PR #79 appears to address one SSRF class: hos...

SSRF Agentic / MCP mcp server tool use
6.3
CVSS
1 month ago