instances
Pass
Audited by Gen Agent Trust Hub on Mar 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill enables direct command execution on remote Koyeb instances using the
koyeb instances execcommand. This allows the agent to run arbitrary code within the instance environment. - [COMMAND_EXECUTION]: The CLI reference provides a complex shell pipeline involving
awk,sort, andxargsto perform bulk deletion of applications (koyeb apps delete), which is a high-impact destructive operation. - [DATA_EXFILTRATION]: The skill facilitates reading remote logs (
koyeb instances logs) and copying files between instances and the local system (koyeb instances cp), which involves accessing and moving potentially sensitive data. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection.
- Ingestion points: Malicious instructions could be embedded in instance logs or the output of
execcommands (SKILL.md, references/koyeb-instances-flags.md). - Boundary markers: None are present in the instructions to help the agent distinguish between data and instructions.
- Capability inventory: The skill has high-privilege capabilities including remote code execution and file system access.
- Sanitization: There is no evidence of output sanitization or validation of the content retrieved from the instances.
Audit Metadata