cekura-onboarding
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests and processes untrusted external data.
- Ingestion points: The skill prompts users to provide their full agent system prompts in 'phase2-agent.md' and call transcripts in 'phase3-observability-ingest.md'.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the ingested text are present.
- Capability inventory: Ingested content is used to drive automated evaluator generation ('scenarios_generate_bg') and metric evaluation ('call_logs_evaluate_metrics_create').
- Sanitization: The instructions do not specify any sanitization or validation routines for the natural language content provided in these fields.
- [SAFE]: The skill adheres to security best practices for a vendor-authored configuration tool.
- The skill uses official vendor domains ('cekura.ai') and identifies its own GitHub organization ('cekura-ai') for plugin installation.
- Authentication guidance refers users to official OAuth flows and standard API key headers without hardcoded secrets or exposure to third-party domains.
- The use of platform-specific tools like 'aiagents_create' and 'scenarios_run_voice' is consistent with the skill's stated purpose of platform onboarding.
Audit Metadata