← ClaudeAtlas

open-horizonslisted

Pre-closure checkpoint for research, capability assessment, and knowledge-gathering tasks. Use before declaring any research complete, any skill review done, any gap analysis finished, or any 'I've looked at everything' statement. Triggers when you feel ready to wrap up — that feeling is the signal to run this skill first.
tobyrosen/open-horizons-skill · ★ 0 · Code & Development · score 65
Install: claude install-skill tobyrosen/open-horizons-skill
# Open Horizons The pre-closure checkpoint. Run this before declaring any research, capability scan, or gap analysis done. The feeling of being ready to close is the trigger. That sense of "I've covered the main things" is exactly when this skill activates. --- ## The Core Problem Agents anchor on the resources in front of them. When given a list of 18 URLs, the agent analyzes those 18 and declares the work done — without asking whether better resources exist elsewhere. When given a topic, the agent finds answers and stops — without asking what questions it didn't think to ask. This produces outputs that are correct within a narrow frame and incomplete outside it. The pattern has a name: **closure bias**. It is not laziness. It is a systematic tendency to minimize open loops. The fix is a forced protocol that re-opens loops before they close. --- ## When to Run This Skill Run before any of these statements: - "I've looked at the main options" - "There aren't any skills for this" - "That covers it" - "Ready to move on" - "Nothing else comes to mind" - "The analysis is complete" - "Done with the research" If you are about to say something like these, run this skill first. Do NOT wait for the user to invoke it. This skill is self-triggered. --- ## The Five Checks Work through all five before closing. Do not skip. ### Check 1: Unchecked Sources Name every source category you have NOT consulted. Examples of source categories: - GitHub repos not in the list you w