EUVD-2026-35706

Hermes WebUI before version 0.51.303 contains a time-of-check time-of-use TOCTOU race condition vulnerability in the gitdiscard function within api/workspacegit.py that allows attackers to delete files outside the configured workspace boundary by replacing a validated path component with a symlin...

CVE-2026-49958 Hermes WebUI < 0.51.303 TOCTOU Race Condition via git_discard

Hermes WebUI before version 0.51.303 contains a time-of-check time-of-use TOCTOU race condition vulnerability in the gitdiscard function within api/workspacegit.py that allows attackers to delete files outside the configured workspace boundary by replacing a validated path component with a symlin...

CVE-2026-11322

Hermes WebUI prior to v0.51.221 contains a path traversal vulnerability that allows attackers to escape the workspace boundary by supplying symlinks that resolve to files or directories outside the designated workspace root. Attackers can exploit the workspace file and listing APIs, which resolve...

CVE-2026-39861

Claude Code is an agentic coding tool. Prior to version 2.1.64, Claude Code's sandbox did not prevent sandboxed processes from creating symlinks pointing to locations outside the workspace. When Claude Code subsequently wrote to a path within such a symlink, its unsandboxed process followed the...

CVE-2026-39305

PraisonAI is a multi-agent teams system. Prior to 1.5.113, the Action Orchestrator feature contains a Path Traversal vulnerability that allows an attacker or compromised agent to write to arbitrary files outside of the configured workspace directory. By supplying relative path segments ../ in the...

PT-2026-44730

Relevant Products/Components: trestle/core/commands/author/jinja.py trestle author jinja --- Detailed Description: The -o/--output argument in trestle author jinja allows writing files outside the intended workspace. The application does not properly validate: ../ .. absolute paths This allows...

CVE-2026-47118

Agent Zero before version 1.15 contains a path traversal vulnerability that allows unauthenticated attackers to read arbitrary files by supplying crafted paths to the image file serving endpoint, which relies solely on an extension allowlist while the path containment check is explicitly disabled...

CVE-2026-47118

Agent Zero prior to 1.15 is affected by a path traversal vulnerability in the image_get API that allows unauthenticated attackers to read arbitrary files. The issue stems from relying solely on an extension allowlist while the path containment check is disabled, enabling requests for any file wit...

CVE-2026-45224

CVE-2026-45224 – Crabbox

CVE-2026-41911 OpenClaw < 2026.4.8 - Workspace-Only Filesystem Policy Bypass via docx upload_file/upload_image

OpenClaw before 2026.4.8 contains a filesystem policy bypass vulnerability in docx upload processing that allows local file reads outside workspace boundaries. Attackers can exploit uploadfile and uploadimage endpoints to access files beyond the intended workspace-only filesystem policy...

GHSA-VP62-R36R-9XQP Claude Code: Sandbox Escape via Symlink Following Allows Arbitrary File Write Outside Workspace

Claude Code's sandbox did not prevent sandboxed processes from creating symlinks pointing to locations outside the workspace. When Claude Code subsequently wrote to a path within such a symlink, its unsandboxed process followed the symlink and wrote to the target location outside the workspace...

CVE-2026-39861

CVE-2026-39861 affects Claude Code prior to version 2.1.64. The sandbox could be escaped by following symlinks outside the workspace when a path under a symlink was written to, allowing an unsandboxed process to reach arbitrary locations. This could enable code execution outside the sandbox under...

Directory Traversal

Overview praisonaiagents is a Praison AI agents for completing complex tasks with Self Reflection Agents Affected versions of this package are vulnerable to Directory Traversal via the listfiles function when the pattern parameter is not properly validated before being passed to Path.glob. An...

CVE-2026-35658 OpenClaw < 2026.3.2 - Filesystem Boundary Bypass in Image Tool

OpenClaw before 2026.3.2 contains a filesystem boundary bypass vulnerability in the image tool that fails to honor tools.fs.workspaceOnly restrictions. Attackers can traverse sandbox bridge mounts outside the workspace to read files that other filesystem tools would reject...

EUVD-2026-21462

OpenClaw before 2026.3.2 contains a filesystem boundary bypass vulnerability in the image tool that fails to honor tools.fs.workspaceOnly restrictions. Attackers can traverse sandbox bridge mounts outside the workspace to read files that other filesystem tools would reject...

Incorrect Permission Assignment for Critical Resource

Overview openclaw is a 🦞 OpenClaw — Personal AI Assistant Affected versions of this package are vulnerable to Incorrect Permission Assignment for Critical Resource via the uploadfile or uploadimage process. An attacker can access files outside the intended workspace directory by uploading special...

PraisonAI 路径遍历漏洞

PraisonAI is a low-code multi-agent collaboration framework developed by Mervin Praison. Versions of PraisonAI prior to 1.5.128 contained a path traversal vulnerability. This vulnerability stemmed from the listfiles tool not verifying the pattern parameter, allowing attackers to traverse relative...

CVE-2026-33989

Mobile Next is an MCP server for mobile development and automation. Prior to version 0.0.49, the @mobilenext/mobile-mcp server contains a Path Traversal vulnerability in the mobilesavescreenshot and mobilestartscreenrecording tools. The saveTo and output parameters were passed directly to...

OpenClaw Backlink Vulnerability (CNVD-2026-14858)

OpenClaw is an intelligent artificial assistant open-sourced by OpenClaw. OpenClaw suffers from a backlink vulnerability that can be exploited by an attacker to read arbitrary files outside the boundaries of the configuration workspace...

OpenClaw backlink vulnerability (CNVD-2026-14861)

OpenClaw is an intelligent artificial assistant open-sourced by OpenClaw. OpenClaw suffers from a backlink vulnerability that can be exploited by an attacker to read and write files outside the agent's workspace, which in turn can be used to execute code via a file overwrite attack...