llm-wiki-setup

An agent skill by daymade, from daymade/claude-code-skills. Tags: finance, knowledge-base, markdown, planning, research.

What it does

Co-create a personal investment-research LLM Wiki (Andrej Karpathy's pattern) where the user's OWN analysis framework becomes a living CLAUDE.md — by interviewing them, NOT by handing them a template. Use whenever the user wants to build a compounding research knowledge base, 投研第二大脑, 投研知识库, or 个人投研 wiki; instantiate Karpathy's LLM Wiki gist for finance/investing; turn their stock-picking, analyst-tracking, or earnings-watching workflow into a structured markdown vault; or build a wiki tracking companies / industries / macro / analysts over time. Pure markdown + wikilinks; no RAG in place of compilation (Karpathy's core idea — qmd search over the grown wiki is in-pattern). Also triggers for ingesting research reports / earnings calls / expert notes into an existing wiki, and for post-earnings prediction→fulfillment reviews. Core value = extracting the user's personal investment preferences into THEIR OWN schema, never imposing a standard one.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add daymade/claude-code-skills --skill llm-wiki-setup

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/daymade/claude-code-skills
cp -r claude-code-skills/llm-wiki-setup ~/.claude/skills/llm-wiki-setup

Safety box score

Grade B, low risk: 23 of 100 risk points, rated 2026-10-16. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill scaffolds a local, markdown-based investment wiki and installs a repo-local git pre-commit lint hook. The main risks are local file writes/overwrites and automatic git configuration changes; it does not appear to fetch remote code, touch secrets, or perform networked actions.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	It ships executable Python tooling (`scripts/init_vault.py`, `scripts/lint-vault.py`) and expects shell execution of those scripts, but it does not eval downloaded code or spawn remote installers.
Network access	0 (none)	No file contains HTTP requests, uploads, or remote downloads; the only remote reference is a Karpathy gist in prose, not an active fetch.
Filesystem reach	2 (moderate)	`init_vault.py` copies an entire vault scaffold into a chosen target and `--refresh-tools` overwrites managed files with `.before-refresh` backups, so it can materially change files in the selected directory tree.
Secrets & credentials	0 (none)	Nothing reads `.env`, keychains, browser sessions, SSH keys, or asks for tokens/passwords.
Privilege & persistence	1 (low)	It requires repo-local git configuration changes such as `git config core.hooksPath .githooks`, which alters commit behavior, but it does not request sudo or global system settings.
Instruction hijack surface	0 (none)	There is no fetched prompt content, hidden instruction layer, or command to ignore higher-priority rules; the 'don't copy templates' wording is ordinary workflow guidance.
Irreversible actions	1 (low)	The local pre-commit hook is designed to automatically block commits on lint failures, but the skill does not publish, deploy, message, pay, or delete anything externally.
Transparency	0 (none)	All shipped logic is plain-text Markdown/Python; there are no binaries, base64 blobs, or unpinned installer downloads.

Before installing

Run `scripts/init_vault.py` only in an empty scratch directory first.
Read `scripts/init_vault.py` and `scripts/lint-vault.py` to confirm which files it can overwrite and where `.before-refresh` backups go.
Decide whether you want repo-local commit blocking via `git config core.hooksPath .githooks`.
If you expect YAML linting, verify `pyyaml` is installed; otherwise YAML validation is skipped.

Rated 2026-10-16 by an automated review of 11 files from the skill's repository.

Source

Repository: daymade/claude-code-skills (all skills from this repository)
Path: llm-wiki-setup/SKILL.md
Branch: main
Updated: 2026-09-20

content-strategy — When the user wants to plan a content strategy, decide what content to create, or figure out what topics to cover.
marketing-ideas — When the user needs marketing ideas, inspiration, or strategies for their SaaS or software product.
public-relations — When the user wants help with public relations, earned media, press coverage, journalist outreach, or media strategy (not pull requests).
research — Investigate a question against high-trust primary sources and capture the findings as a Markdown file in the repo.
architect — Sketch types, signatures, and module structure before code, then stay in the loop while implementation fills in.
co-marketing — When the user wants to find co-marketing partners, plan joint campaigns, or brainstorm partnership opportunities.