routeros-cmr
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for configuring alert rules that ingest RouterOS system logs to trigger HTTP webhooks. This architecture allows external system logs to enter the agent context and potentially influence downstream actions.
- Ingestion points: System log monitoring and regex matching configured in
SKILL.md. - Boundary markers: The configurations do not specify markers to delimit log data from instructions.
- Capability inventory: Network operations via HTTP webhooks (
action.http-url) and RouterOS CLI command execution as described inSKILL.md. - Sanitization: No explicit sanitization or filtering of log data before alert processing is provided.
- [EXTERNAL_DOWNLOADS]: The skill contains references to external code examples and reproduction reports hosted in the author's repository.
- Evidence: Links to
https://github.com/tikoci/quickchr/tree/main/examples/cmrfound inSKILL.md.
Audit Metadata