Store never-do rules with NEVER, BECAUSE, and INSTEAD structures.
npx skills add https://github.com/Nerfherder16/BrickLayer --skill never
Please help me install this Agent Skill. Skill: never Source: https://github.com/Nerfherder16/BrickLayer/tree/main/.claude/skills/never Command: npx skills add https://github.com/Nerfherder16/BrickLayer --skill never
Stores an anti-pattern that records a standing never-do rule to prevent specific actions in Claude workflows and interactions.
Add a never rule specifying the forbidden action, the reason, and the recommended alternative so Claude permanently avoids the action.
High-intent search queries and answers about installing and using this skill.
Persistent guardrails prevent unsafe AI actions by storing explicit never-do rules as anti-patterns. You define a forbidden action, the reason, and an alternative to ensure consistent enforcement across Claude Code sessions.
An anti-pattern for enforcing AI prohibitions is a structured never-do rule that prevents specific actions like revealing secrets. You need it when forward-looking guardrails must permanently guide responses and stop unsafe behaviors.
Format a never-do rule using the NEVER, BECAUSE, INSTEAD structure. Specify the forbidden action, explain the reason, and provide a recommended alternative so Claude permanently avoids the action.
Yes, standing prohibition rules apply across Claude Code sessions by persisting anti-patterns. These stored prohibitions are consulted before decisions to capture guardrails that should never be triggered.
Enforcing absolute prohibitions requires no specific dependencies or components. It relies on a structured template to capture, store, and recall rules, ensuring forward-looking guardrails guide AI behavior.
Structured prohibition templates provide explicit NEVER, BECAUSE, INSTEAD formatting to guide responses, unlike standard instructions. This ensures persistent guardrails are consistently consulted before decisions to prevent prohibited actions.
OzeroHAXLoads mandatory baseline rules before agent task execution.
shynlee04Detect output style deviations and insert non-blocking reminders in iDumb agent responses.
Owl-ListenerDefine behavioral guardrails for AI content, actions, tone, scope, and confidence.
CAPHTECHDesign evaluation tests probing failure modes in LLM-based systems.
