render-background-workers
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of background workers that ingest and process data from external queues (Key Value/Redis, PostgreSQL). This represents a vulnerability surface where untrusted job payloads enter the agent's context.\n
- Ingestion points: Worker implementations in
references/queue-framework-setup.mdpoll jobs from external data stores (Key Value, Postgres).\n - Boundary markers: The provided templates do not include instructions or markers to delimit or ignore instructions within the processed job data.\n
- Capability inventory: The skill configures service execution via
startCommandandbuildCommandas seen inSKILL.md.\n - Sanitization: There is no evidence of data sanitization or validation logic for queue-ingested content in the snippets.\n- [COMMAND_EXECUTION]: The skill instructions and reference files contain shell commands intended for platform configuration and service management.\n
- Evidence:
SKILL.mdandreferences/queue-framework-setup.mddefinestartCommand(e.g.,celery -A tasks worker,bundle exec sidekiq,node worker.js) andbuildCommand(e.g.,pip install -r requirements.txt) for various runtimes.
Audit Metadata