CodeWhale before 0.8.64 Privilege Escalation via exec_shell_interact
Hmbown CodeWhale
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| deepseek-tui crates.io |
| No fix yet |
| deepseek-tui npm |
|
|
| codewhale-tui crates.io |
|
Hmbown CodeWhale
| Product | Affected | Fixed in |
|---|---|---|
| deepseek-tui crates.io |
| No fix yet |
| deepseek-tui npm |
|
|
| codewhale-tui crates.io |
|
|
| codewhale npm |
|
|
CodeWhale versions >= 0.3.10 and < 0.8.64 (published as deepseek-tui >= 0.3.10 and < 0.8.41 on npm and 0.3.10 through 0.8.41 on crates.io, and as codewhale / codewhale-tui >= 0.8.41 and < 0.8.64) contain a vulnerability in the exec_shell_interact (alias exec_interact) tool, whose approval_requirement returns ApprovalRequirement::Auto. This overrides the default Required approval for code-executing tools, so LLM-controlled stdin is written into an already-approved long-running interactive shell (e.g., a python3 -i REPL, mysql, ssh, or sudo -i session) without any approval prompt. An attacker who can inject instructions via untrusted content the agent ingests (a fetched page, MCP result, or repo file) can cause commands to run at the privilege level of that approved process. Fixed in codewhale / codewhale-tui 0.8.64; the deprecated deepseek-tui crate has no fixed release.
From the CVE record · VulnCheck
CodeWhale’s exec_shell_interact tool can write LLM-controlled input to an already-approved interactive shell without another approval prompt. An attacker who injects instructions into content the agent ingests can cause commands to run with that shell process’s privileges; this requires an approved session and the agent to ingest the attacker-controlled content.
Check installations of npm deepseek-tui >=0.3.10 <0.8.41, crates.io deepseek-tui >=0.3.10 <=0.8.41, npm codewhale >=0.8.41 <0.8.64, and crates.io codewhale-tui >=0.8.41 <0.8.64.