← ClaudeAtlas

event-treelisted

Organize large real-world events into a queryable tree (_INDEX + topic nodes). Top-down retrieval, full detail retention, decision log with re-review triggers. Use when an event's context exceeds one file.
ZongrongLi/event-tree-skill · ★ 0 · AI & Automation · score 65
Install: claude install-skill ZongrongLi/event-tree-skill
# Event Tree Turn any large, ongoing real-world event (a legal case, a job negotiation, a house hunt, a long-term plan) into a tree of Markdown files an agent can navigate top-down. Built for decisions, not just storage. ## Why a tree? A single topic file past ~15KB becomes an "information wall": the agent can't read it fully, can't locate facts fast, and burns tokens on every recall. A tree fixes three things: 1. **Top-down indexing** — read `_INDEX.md` first (status snapshot + open decision points + node map), then drill into detail files only on demand. Never load the whole tree. 2. **Bounded file size** — each node stays ≤ ~8KB, so one read call covers it completely. 3. **Separation of detail from decision** — facts (timeline, evidence, quotes) and decisions (options, trade-offs, triggers) live in different files and never contaminate each other. ## Storage layout ``` <event>/ ├── _INDEX.md # Tree root: status snapshot + open decision points + node map (always read this first) ├── 01-timeline.md # Chronological event log (quotes, numbers, dates preserved verbatim) ├── 02-stakeholders.md # People/orgs: positions, interests, relationships ├── 03-evidence.md # File paths, original quotes, links, screenshots ├── 04-analysis.md # Agent deep-dive: legal/strategic/game-theoretic analysis ├── 05-decision-log.md # Every decision: options, trade-offs, choice, re-review trigger ├── 06-actions.md # Todos, next steps, deadlines └── 07-risks.md #