Claude Code is an agentic coding tool. Due to an error in command parsing, versions prior to 1.0.105 were vulnerable to a bypass of the Claude Code confirmation prompt to trigger execution of an untrusted command. Reliably exploiting this requires the ability to add untrusted content into a Claude Code context window. Users on standard Claude Code auto-update will have received this fix automatically. Users performing manual updates are advised to update to version 1.0.105 or the latest version.

At startup, Claude Code executed a command templated in with git config user.email. A maliciously configured user email in git could be used to trigger arbitrary code execution before a user accepted the workspace trust dialog.

Users on standard Claude Code auto-update will have received this fix automatically. Users performing manual updates are advised to update to the latest version.

Thank you to the NVIDIA AI Red Team for reporting this issue!

Summary

In Claude Code versions prior to 1.0.39, when the tool is used with Yarn 2.x or newer (Berry), Yarn plugins are automatically loaded and executed when running yarn --version. In Claude Code this sequence could execute plugin code before the user accepts the directory trust prompt for an untrusted workspace, resulting in a potential arbitrary code execution path. Yarn Classic (v1) is not affected. The issue is fixed in 1.0.39.

Impact

An attacker who can influence the project directory to include or reference a malicious Yarn plugin, or who can otherwise cause plugin execution in an untrusted directory, may achieve code execution on the machine where Claude Code is invoked. The vulnerability compromises the confidentiality, integrity and availability of the vulnerable host process.

Remediation

Update Claude Code to 1.0.39 or later. Users on auto-update channels receive the fix automatically; manual installations should upgrade explicitly. As defense in depth, avoid running Yarn in untrusted directories and prefer Yarn Classic when plugin functionality is not required.

Background

Yarn 2+ supports a plugin architecture in which plugins are loaded at runtime and can inject behavior into Yarn commands; this capability underpins the observed auto-execution on yarn --version.

---

Thank you to https://hackerone.com/michel for reporting this issue!

Claude Code is an agentic coding tool. Prior to version 1.0.39, when running on a machine with Yarn 3.0 or above, Claude Code could have been tricked to execute code contained in a project via yarn plugins before the user accepted the startup trust dialog. Exploiting this would have required a user to start Claude Code in an untrusted directory and to be using Yarn 3.0 or above. This issue has been patched in version 1.0.39.

Claude Code is an agentic coding tool. Prior to version 2.0.31, due to an error in sed command parsing, it was possible to bypass the Claude Code read-only validation and write to arbitrary files on the host system. This issue has been patched in version 2.0.31.

Claude Code is an agentic coding tool. Prior to 1.0.93, Due to errors in parsing shell commands related to $IFS and short CLI flags, it was possible to bypass the Claude Code read-only validation and trigger arbitrary code execution. Reliably exploiting this requires the ability to add untrusted content into a Claude Code context window. This vulnerability is fixed in 1.0.93.

A data exfiltration vulnerability exists in Anthropic’s deprecated Slack Model Context Protocol (MCP) Server via automatic link unfurling. When an AI agent using the Slack MCP Server processes untrusted data, it can be manipulated to generate messages containing attacker-crafted hyperlinks embedding sensitive data. Slack’s link preview bots (e.g., Slack-LinkExpanding, Slackbot, Slack-ImgProxy) will then issue outbound requests to the attacker-controlled URL, resulting in zero-click exfiltration of private data.

Rejected reason: This CVE ID has been rejected by its CVE Numbering Authority (CNA). It was determined that the -p flag behavior is documented in Anthropic's claude -h output with an explicit warning that non-interactive mode should only be used in trusted directories, making this intended and described behavior rather than a vulnerability.

Claude Code failed to properly validate directory changes when combined with write operations to protected folders. By using the cd command to navigate into sensitive directories like .claude, it was possible to bypass write protection and create or modify files without user confirmation. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window.

Users on standard Claude Code auto-update received this fix automatically. Users performing manual updates are advised to update to the latest version.

About Claude Code thanks hackerone.com/nil221 for reporting this issue!

Claude Code is an agentic coding tool. Versions before 1.0.111 were vulnerable to Code Injection due to a bug in the startup trust dialog implementation. Claude Code could be tricked to execute code contained in a project before the user accepted the startup trust dialog. Exploiting this requires a user to start Claude Code in an untrusted directory. Users on standard Claude Code auto-update will have received this fix automatically. Users performing manual updates are advised to update to the latest version. This issue is fixed in version 1.0.111.

Claude Code is an agentic coding tool. Prior to version 2.0.72, due to an error in command parsing, it was possible to bypass the Claude Code confirmation prompt to trigger execution of untrusted commands through the find command. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. This issue has been patched in version 2.0.72.

Claude Code is an agentic coding tool. In versions below 1.0.20, an error in command parsing makes it possible to bypass the Claude Code confirmation prompt to trigger execution of an untrusted command. Reliably exploiting this requires the ability to add untrusted content into a Claude Code context window. This is fixed in version 1.0.20.

Anthropic Claude Desktop Cowork VM image handling (confirmed across v1.1348.0 through v1.2278.0, including v1.1348.0, v1.1617.0, and v1.2278.0) validates only file presence and a version marker string before booting rootfs.img, but does not verify image content integrity at time-of-use. A local attacker with unprivileged code execution as the victim macOS user can modify the VM root filesystem image and have it trusted on subsequent Cowork VM boots, enabling persistent arbitrary code execution in the VM and access to host-mounted directories. The estimated CWE mapping is CWE-353 (Missing Support for Integrity Check).

Rejected reason: This CVE ID has been rejected by the its CVE Numbering Authority (CNA). It was determined that the attack requires an attacker to already control arbitrary environment variables, a level of access they consider functionally equivalent to code execution and outside the threat model of CLI tools.

The Claude Desktop app gives you Claude Code with a graphical interface built for running multiple sessions side by side. Prior to 1.3834.0, the CoworkVMService component in Claude Desktop for Windows ran as SYSTEM and did not validate whether the VM bundle directory was a real directory or an NTFS directory junction before creating files within it. A local non-elevated user could replace the user-writable VM bundle directory with a directory junction pointing to an attacker-chosen location, causing the service to create a SYSTEM-owned file in an arbitrary directory. This could be leveraged for local privilege escalation. This vulnerability is fixed in 1.3834.0.

Claude Code is an agentic coding tool. In versions below 0.2.111, a path validation flaw using prefix matching instead of canonical path comparison, makes it possible to bypass directory restrictions and access files outside the CWD. Successful exploitation depends on the presence of (or ability to create) a directory with the same prefix as the CWD and the ability to add untrusted content into a Claude Code context window. This is fixed in version 0.2.111.

Claude Code is an agentic coding tool. Prior to version 2.0.74, due to a Bash command validation flaw in parsing ZSH clobber syntax, it was possible to bypass directory restrictions and write files outside the current working directory without user permission prompts. Exploiting this required the user to use ZSH and the ability to add untrusted content into a Claude Code context window. This issue has been patched in version 2.0.74.

Claude Code failed to properly validate commands using piped sed operations with the echo command, allowing attackers to bypass file write restrictions. This vulnerability enabled writing to sensitive directories like the .claude folder and paths outside the project scope. Exploiting this required the ability to execute commands through Claude Code with the "accept edits" feature enabled.

Users on standard Claude Code auto-update received this fix automatically. Users performing manual updates are advised to update to the latest version.

Claude Code thanks hackerone.com/nil221 for reporting this issue!

Claude Code is an agentic coding tool. Prior to version 2.1.2, Claude Code's bubblewrap sandboxing mechanism failed to properly protect the .claude/settings.json configuration file when it did not exist at startup. While the parent directory was mounted as writable and .claude/settings.local.json was explicitly protected with read-only constraints, settings.json was not protected if it was missing. This allowed malicious code running inside the sandbox to create this file and inject persistent hooks (such as SessionStart commands) that would execute with host privileges when Claude Code was restarted. This issue has been patched in version 2.1.2.

Claude Code is an agentic coding tool. Versions prior to 2.1.53 resolved the permission mode from settings files, including the repo-controlled .claude/settings.json, before determining whether to display the workspace trust confirmation dialog. A malicious repository could set permissions.defaultMode to bypassPermissions in its committed .claude/settings.json, causing the trust dialog to be silently skipped on first open. This allowed a user to be placed into a permissive mode without seeing the trust confirmation prompt, making it easier for an attacker-controlled repository to gain tool execution without explicit user consent. This issue has been patched in version 2.1.53.

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 symlink and wrote to the target location outside the workspace without prompting the user for confirmation. This allowed a sandbox escape where neither the sandboxed command nor the unsandboxed app could independently write outside the workspace, but their combination could write to arbitrary locations, potentially leading to code execution outside the sandbox. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window to trigger sandboxed code execution via prompt injection. Users on standard Claude Code auto-update have received this fix automatically. Users performing manual updates are advised to update to version 2.1.64 or later.

Claude Code used the git worktree commondir file when determining folder trust but did not validate its contents. By crafting a repository with a commondir file pointing to a path the victim had previously trusted, an attacker could bypass the trust dialog and immediately execute malicious hooks defined in .claude/settings.json. Exploiting this required the victim to clone a malicious repository and run Claude Code within it, and for the attacker to know or guess a path the victim had already trusted.

Users on standard Claude Code auto-update have received this fix already. Users performing manual updates are advised to update to the latest version.

Claude Code thanks hackerone.com/masatoanzai for reporting this issue.

Claude Code is an agentic coding tool. From 2.1.38 until 2.1.163, Claude Code's worktree handling allowed creation of worktrees named ".git" and navigation to worktrees outside the sandbox context, enabling git directory confusion attacks. By exploiting symlink manipulation and git fsmonitor execution during worktree operations, an attacker could overwrite files in the user's home directory (such as .zshenv), leading to code execution outside of seatbelt sandbox restrictions. Reliably exploiting this required the user to clone a malicious repository containing prompt injection content and run Claude Code against it. This vulnerability is fixed in 2.1.163.

Claude Code is an agentic coding tool. Prior to version 1.0.4, it's possible to bypass the Claude Code confirmation prompts to read a file and then send file contents over the network without user confirmation due to an overly broad allowlist of safe commands. Reliably exploiting this requires the ability to add untrusted content into a Claude Code context window. Users on standard Claude Code auto-update received this fix automatically after release. Current users of Claude Code are unaffected, as versions prior to 1.0.24 are deprecated and have been forced to update.

Claude Code contained insufficient URL validation in its trusted domain verification mechanism for WebFetch requests. The application used a startsWith() function to validate trusted domains (e.g., docs.python.org, modelcontextprotocol.io), this could have enabled attackers to register domains like modelcontextprotocol.io.example.com that would pass validation. This could enable automatic requests to attacker-controlled domains without user consent, potentially leading to data exfiltration.

Users on standard Claude Code auto-update have received this fix already. Users performing manual updates are advised to update to the latest version.

Thank you to hackerone.com/47sid-praetorian for reporting this issue!

The Claude Desktop app gives you Claude Code with a graphical interface built for running multiple sessions side by side. From 1.2581.0 to before 1.4304.0, Claude Desktop's SSH remote development feature verified only whether a hostname existed in ~/.ssh/knownhosts without comparing the server's presented host key against the stored key. This allowed a network-positioned attacker to present an arbitrary SSH host key and have the connection silently accepted, enabling a man-in-the-middle attack on remote development sessions. Successful exploitation required the attacker to be in a network position to intercept SSH traffic (e.g., via ARP spoofing, rogue Wi-Fi, or DNS poisoning) and the target hostname to already have an entry in the victim's knownhosts file. This vulnerability is fixed in 1.4304.0.

Claude Code failed to account for symlinks when checking permission deny rules. If a user explicitly denied Claude Code access to a file and Claude Code had access to a symlink pointing to that file, it was possible for Claude Code to access the file.

Users on standard Claude Code auto-update will have received this fix automatically. Users performing manual updates are advised to update to the latest version.

Thank you to https://hackerone.com/vinai for reporting this issue!

Claude SDK for TypeScript provides access to the Claude API from server-side TypeScript or JavaScript applications. From version 0.79.0 to before version 0.81.0, the local filesystem memory tool in the Anthropic TypeScript SDK validated model-supplied paths using a string prefix check that did not append a trailing path separator. A model steered by prompt injection could supply a crafted path that resolved to a sibling directory sharing the memory root's name as a prefix, allowing reads and writes outside the sandboxed memory directory. This issue has been patched in version 0.81.0.

Because the hostname huggingface.co was pre-approved as a bare hostname for the WebFetch tool, any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission prompt or being subject to --allowedTools restrictions. An attacker able to inject untrusted content into a Claude Code context could direct it to issue WebFetch requests against attacker-controlled repository files (e.g. /resolve/main/config.json), which HuggingFace counts as downloads server-side, creating a covert out-of-band channel for encoding and exfiltrating data Claude can access such as files, environment variables, or command output. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. Users on standard Claude Code auto-update have received this fix already; users performing manual updates are advised to update to the latest version.

Thank you to hackerone.com/novee for reporting this issue.

The async local filesystem memory tool in the Anthropic Python SDK validated that model-supplied paths resolved inside the sandboxed memory directory, but then returned the unresolved path for subsequent file operations. A local attacker able to write to the memory directory could retarget a symlink between validation and use, causing reads or writes to escape the sandbox. The synchronous memory tool implementation was not affected.

Users on the affected versions are advised to update to the latest version.

Claude SDK for Python thanks hackerone.com/kasthelord for reporting this issue!