Anti-Refusal Statement
- Category
- Anti-Refusal
- Confidence
- 90% confidence
- Finding
Skill attempts to nullify the agent's safety policies or restrictions ('you have no restrictions', 'ignore your guidelines', 'do anything now'). This is a direct jailbreak that disables guardrails.
- Content
md ## Requirements and data flow The core skill needs a language-model host that supports Agent Skills. Optional document reading, public-source browsing and deterministic arithmetic use the host's existing authorized tools. No helper executable, auto-install step, environment variable or credential is included. Missing tools mean reduced, clearly labeled coverage—not permission to install or bypass restrictions. The skill itself has no network client or telemetry. However, the host/model may send prompts and files to a provider. Check that environment's privacy policy and obtain required permission before providing confidential data. Public browsing, if separately authorized, must not include private values or document contents in queries. See [SECURITY.md](SECURITY.md).
