About this tracker

What this is

A public, moderated database of behavioral patterns observed across AI systems. Each entry describes what was seen, where, and with what evidence. Anyone can submit. Admins review before publishing.

Why a database and not an essay

Patterns become legible only when examples accumulate. A single screenshot is anecdote; fifty entries tagged Premature Closure across three model families is a pattern. The goal here is to give people a place to log what they notice so the field has something to look at.

Moderation and history

Admins can publish, edit, or reject submissions. Nothing is silently changed — every edit writes a row to the entry's history with a diff of what changed and when. Rejections carry a reason that stays on the record.

Relationship to Consciera

This tracker is a companion to the research context at consciera.ai/pattern-documentation. The main site documents patterns observed within a specific inquiry. This wiki is open — a place for the wider community to contribute observations that may or may not match those patterns.

What submissions should contain

• A clear title and description of the behavior — specific enough that another reader could recognize it.
• Which system and model version. If the model updated between observations, note that.
• Prompt and context when possible. "First message" vs. "deep in a long conversation" matters.
• Evidence: screenshots, a text excerpt, or a transcript PDF. 10 MB max per file.

Contact

Comments on moderation, requests to remove an entry, or questions about the framing: reach out through consciera.ai.