LLM Wiki Setup (Investment Research Second Brain)

What it does

• Cleanly separates the mechanism layer (three-tier folders, [[wikilinks]], a lint script, git pre-commit hook — installed automatically via scripts/init_vault.py) from the rules layer (what you actually look at and how you record judgments).
• The core move is an interview, not a template. It walks through 8 dimensions (tracked metrics, how you log a thesis, whether to attribute views to analysts, how you review after earnings, note length, etc.) and writes the rules layer into CLAUDE.md in your own words.
• Deliberately no RAG, vector DB, or embeddings. Knowledge is pre-compiled into structured pages so it compounds, instead of re-retrieving raw documents on every query — that's the whole point of the pattern.
• Ships SOPs for ingesting sources (5 human-in-the-loop checkpoints) such as sell-side reports, earnings calls and expert notes; a prediction→fulfillment review SOP for post-earnings; plus pruning discipline and counter-review guides.

Who it's for

• Investors tracking companies, industries, macro and analysts over years who want their reasoning to accumulate.
• Analysts and researchers whose reports and call notes pile up unused in a folder.
• Fans of Obsidian-style markdown vaults who don't want heavyweight retrieval infrastructure.
• Anyone who wants to document their own investment brain rather than adopt someone else's standard schema.

Examples

1. New vault: ask for an "investment research second brain" — it scaffolds the empty skeleton, then interviews you and records your real criteria (e.g. "do next-quarter orders beat consensus", "management tone on the call").
2. Ingest a source: hand it a semiconductor sector report; it splits the content into company / industry / analyst pages wired together with wikilinks.
3. Post-earnings review: after results drop, it compares your logged prediction against the actual outcome, writes a review page, and can hand off to the analyst-track-record skill to backtest an analyst's hit rate.

Install the skill I found on Claude Skill Mart.
Copy the llm-wiki-setup folder from the GitHub repo daymade/claude-code-skills into my ~/.claude/skills/llm-wiki-setup/.
When it's done, tell me in one line what this skill can do.