z-go-messaging
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for building message consumers that ingest data from external brokers (Kafka, RabbitMQ, and NATS), creating a surface for indirect prompt injection. \n
- Ingestion points: Consumer handlers in
SKILL.md(Idempotent consumers) andreferences/brokers.md(Kafka, RabbitMQ, and NATS consume loops) process records from external broker subjects and topics. \n - Boundary markers: The provided code snippets do not implement specific delimiters or 'ignore' instructions for the message payloads. \n
- Capability inventory: The skill demonstrates handlers performing database writes (
tx.ExecinSKILL.md) and network operations via broker clients (producer.Publish,cl.PollFetches). \n - Sanitization: There is no explicit evidence of payload sanitization or validation in the instructional code snippets. \n- [COMMAND_EXECUTION]: The skill includes a shell command for verification of the implementation via Go's testing tool. \n
- Evidence:
timeout 15m go test -parallel 4 -p 2 -race -timeout 10m ./internal/consumer/... ./internal/outbox/...inSKILL.md. This is a standard developer verification workflow.
Audit Metadata