dedaub-monitoring
Fail
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: CRITICALEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill includes links to websites flagged as malicious by automated scanners. These URLs are presented as authoritative documentation sources for the Beamer and EigenLayer protocols in files such as references/protocols/eigenlayer/core.md. Flagged URLs include docs.beamerbridge.com and docs.eigencloud.xyz/eigenlayer/.
- [PROMPT_INJECTION]: The skill's core functionality involves generating and deploying SQL monitoring alerts based on unvalidated free-text user input ("Condition") as described in SKILL.md. The instructions explicitly direct the agent to trust user-supplied "signal identities" (Step 0, section 4a), bypassing local verification. This creates a risk where malicious users could provide SQL fragments that are deployed as part of an alert. Boundary markers and explicit sanitization instructions for these user-provided strings are absent.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute CLI commands using user-provided parameters, which can lead to command injection. For example, Step 0, section 4d of SKILL.md provides a template for creating webhooks: 'dedaub-monitoring webhook create --name --url '. Special characters in the user-controlled or fields could be used to manipulate the command execution.
Recommendations
- AI detected serious security threats
- Contains 4 malicious URL(s) - DO NOT USE
Audit Metadata