observability-debugging
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous commands utilizing
curl,kubectl, and thenaisCLI to fetch diagnostic data. These operations are directed at Nav's internal observability endpoints, such asmimir.nav.cloud.nais.io,loki.nav.cloud.nais.io, andtempo.nav.cloud.nais.io. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes application logs and traces which may contain malicious payloads intended to influence the agent.
- Ingestion points: The skill reads external data from the Loki (
loki.nav.cloud.nais.io) and Tempo (tempo.nav.cloud.nais.io) APIs. - Boundary markers: No specific delimiters or safety instructions are provided to the agent to distinguish between its instructions and the data retrieved from logs.
- Capability inventory: The agent can perform network operations via
curland interact with the Kubernetes API throughkubectland thenaisCLI. - Sanitization: There is no evidence of content validation or sanitization for the retrieved logs or traces before they are presented to the agent for analysis.
Audit Metadata