design-ai-trust-and-failure-states

SKILLWorkflowcommunity
v0.0.0dineshrevunuruMITUpdated 1mo agoSource →

Design the user-facing reliability layer of any AI product: uncertainty and confidence display, provenance/citations, hallucination and sycophancy mitigations, explainability calibrated to stakes, and a named cause + recovery path for EVERY failure — refusal, timeout, content filter, context overflo

Community-submitted skill. Not yet reviewed by the Forge team. Full prompt content may not be available.Request review →
1Repo stars
1Clients
1Formats
1mo agoLast update
Skill
Authordineshrevunuru
Version0.0.0
LicenseMIT
CategoryWorkflow
Formatsskill.md
PromptNot published
Compatibility
Claude✓ Supported
Cursor
Copilot
ChatGPT
Gemini
About

Design the user-facing reliability layer of any AI product: uncertainty and confidence display, provenance/citations, hallucination and sycophancy mitigations, explainability calibrated to stakes, and a named cause + recovery path for EVERY failure — refusal, timeout, content filter, context overflow, mic/STT death, TTS death, dead live-embed. Use when asked 'what happens when the AI is wrong/fail

Keywords
skillclaude