{"skill":{"slug":"probe-first-research","displayName":"Probe First Research","summary":"Probe-first deep research — low-cost snippet reconnaissance before committing to full searches","tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":70,"installsAllTime":0,"installsCurrent":0,"stars":0,"versions":1},"createdAt":1776154191851,"updatedAt":1776154609877},"latestVersion":{"version":"1.0.0","createdAt":1776154191851,"changelog":"probe_first_research v1.0.0 — Initial release\n\n- Introduces a \"probe-first\" deep research workflow that maps the information landscape using search snippets before deeper research.\n- Automatically generates 2–3 reconnaissance queries and synthesizes a Probe Report summarizing information density, terminology, source types, entities, controversy, and other dimensions.\n- Presents structured research planning (framework selection, sub-question decomposition) based on real search signals, not assumptions.\n- Includes user-invocable stop point for plan confirmation and supports dynamic adjustment (depth, multi-agent execution, freshness constraints).\n- Deep search phase strategically fetches and analyzes high-value sources, guided by initial probe findings.","license":"MIT-0"},"metadata":{"os":null,"systems":null},"owner":{"handle":"lkm123","userId":"s1757m9w8w2sga31a9bfcvc2v184vscs","displayName":"lkm123","image":"https://avatars.githubusercontent.com/u/50452258?v=4"},"moderation":null}