Local command execution via unrestricted input_model exec
Published May 16, 2025 · Updated May 16, 2025
Code injection in Defog Introspect 0.1.0 through 0.1.4 allows local users to execute arbitrary commands through the custom-tools test endpoint. The test_custom_tool handler passes the complete input_model string to Python exec(), allowing imports and built-ins to escape its BaseModel-and-Field namespace. Low-privileged local access is required; successful injection runs with the backend service's privileges and can read, modify, or interrupt its data and processes.
Summary
What happened
Code injection in Defog Introspect 0.1.0 through 0.1.4 allows local users to execute arbitrary commands through the custom-tools test endpoint. The test_custom_tool handler passes the complete input_model string to Python exec(), allowing imports and built-ins to escape its BaseModel-and-Field namespace. Low-privileged local access is required; successful injection runs with the backend service's privileges and can read, modify, or interrupt its data and processes.
The record
- CVE
- CVE-2025-4767
- Published
- May 16, 2025
- Updated
- May 16, 2025
- Vendor
- Defog.ai
- Product
- Defog Introspect
- Classifications
- CWE-74, CWE-94, T1059.006
- Attack vector
- local
- Privileges
- authenticated
Timeline
How it unfolded
- May 16, 2025CVE publishedPublication date reported by the CVE source.
- May 16, 2025Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=0.1.0
- Affected versionversion=0.1.1
- Affected versionversion=0.1.2
- Affected versionversion=0.1.3
- Affected versionversion=0.1.4
What conditions does exploitation require?
What is affected?
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
- Remote Code Execution via Unsafe exec in custom-tools test endpointproof of concept · unverified
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo