The idk-lamp is a quiet signal that lights up when AI should not make a decision.
It is not an error or a failure. It is a "boundary signal" to return judgment to humans.
This is not a "manual", but a page to align the implementation order.
• Instantly understand "what the signal is" via Demo
• "Ready to place" via CDN
• "Consistent trigger points" via Gate
• Detailed Philosophy (→ Concept)
• Full Theme Guide (→ Demo)
• Full Specs & Implementation (→ Repos)
Minimal setup to "just try it out". Ready for sites or PoCs.
* Check the idk-lamp repository for the latest specs on state, theme, and variables.
First, choose what fits your product. Compare the lighting atmosphere in the theme list.
idk-lamp is a "signal".
For the signal to work, you first need to define when to light it (conditions for AI not to decide).
• Translating "uncertainty" in prediction/inference/generation into operational decisions
• Defining conditions for deferring judgment (who checks what and when)
• Establishing that "stopping" is a normal operation, not an error
No. idk-lamp is not for indicating "danger", but a signal to return judgment.
It visualizes the "boundary of responsibility" on the UI, rather than replacing alerts.
Because their roles differ.
Demo is for experience, idk-lamp repo is for artifacts, and predictability-gate repo is for operational judgment.
Unifying only the workflow here prevents the concept from breaking.
If unsure, go Demo → Install (here) → Gate.
Making it "placeable" first brings the Gate discussion into reality.