monitoring-expert
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill includes hardcoded default credentials within its configuration examples. In SKILL.md, Alertmanager is configured with 'auth_password: password'. In references/PROMETHEUS.md, the Grafana service environment sets 'GF_SECURITY_ADMIN_PASSWORD=admin'.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability to indirect prompt injection as it is designed to ingest and process potentially untrusted external data. (1) Ingestion points: Log files defined in promtail-config.yml (/var/log/app/*.log) and metrics scraped from application endpoints. (2) Boundary markers: No explicit delimiters or instructions are provided to help the agent distinguish between data and embedded instructions in log lines or metric labels. (3) Capability inventory: The skill is scoped to use Read, Write, Edit, and Bash (including docker, kubectl, and promtool), which could be misused if a malicious instruction is followed. (4) Sanitization: The configuration lacks filtering or validation mechanisms to identify or strip natural language instructions from metrics and logs.
Audit Metadata