strix
Fail
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill executes remote code by piping a script from an external URL directly into the bash shell.
- Evidence:
scripts/install.shcontainscurl -sSL https://strix.ai/install | bash. - Evidence:
SKILL.md(baseline) andreferences/commands.mdalso recommend this unverified execution pattern for installation. - Impact: This pattern bypasses integrity checks and allows the remote server to execute arbitrary commands with the agent's privileges.
- [EXTERNAL_DOWNLOADS]: The skill downloads executable binaries and Docker images from third-party sources.
- Evidence:
scripts/install.shpulls theghcr.io/usestrix/strix-sandbox:0.1.13Docker image and downloads the Strix binary. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted target URLs and instructions, which could contain malicious directives.
- Ingestion points:
scripts/run-scan.shaccepts user-supplied--targetand--instructionarguments. - Boundary markers: No explicit delimiters or guardrail instructions are used to isolate untrusted data.
- Capability inventory: The skill has access to bash execution, Docker, and sensitive LLM API keys.
- Sanitization: Input data is passed to the underlying security tool without validation or sanitization.
- [COMMAND_EXECUTION]: The skill relies on executing shell commands and external binaries to perform vulnerability scans.
- Evidence:
scripts/run-scan.shandscripts/ci-scan.shuseexecto invoke thestrixCLI with parameters.
Recommendations
- HIGH: Downloads and executes remote code from: https://strix.ai/install - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata