Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/ayanbekdos/memoriki/claude-mdgit clone --depth 1 https://github.com/AyanbekDos/memorikiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/ayanbekdos/memoriki/claude-md)<a href="https://agentmods.dev/instructions/ayanbekdos/memoriki/claude-md"><img src="https://agentmods.dev/badge/instructions/ayanbekdos/memoriki/claude-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00621 | $0.00621 |
| Opus 5 | $0.00311 | $0.00311 |
| Sonnet 5 | $0.00124 | $0.00124 |
| Haiku 4.5 | $0.00062 | $0.00062 |
Grade A, and why
memoriki CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
LLM Wiki - Schema
Overview
This is a personal LLM-powered knowledge base following Andrej Karpathy's "LLM Wiki" pattern. The LLM maintains a structured, interlinked markdown wiki compiled from raw source documents.
Directory Structure
llm-wiki/
idea-file.md # Karpathy's original idea file (reference)
CLAUDE.md # This schema file (instructions for the LLM)
raw/ # Immutable source documents (LLM reads, NEVER modifies)
assets/ # Downloaded images
wiki/ # LLM-generated wiki (LLM owns this entirely)
index.md # Content catalog - updated on every ingest
log.md # Chronological activity log (append-only)
entities/ # Pages about specific people, companies, products
concepts/ # Pages about ideas, patterns, frameworks
sources/ # Summary pages for each ingested source
synthesis/ # Cross-cutting analysis, comparisons, insights
Rules
- NEVER modify files in
raw/- they are immutable source of truth - ALWAYS update
wiki/index.mdafter creating or updating any wiki page - ALWAYS append to
wiki/log.mdafter any operation (ingest, query, lint) - Use [[wiki-links]] to connect related pages
- Add YAML frontmatter to every wiki page:
--- title: Page Title type: entity|concept|source|synthesis sources: [list of raw/ files referenced] related: [[linked-pages]] created: YYYY-MM-DD updated: YYYY-MM-DD --- - Log format:
## [YYYY-MM-DD] operation | Description
Operations
Ingest
When user drops a new source into raw/:
- Read the source document
- Discuss key takeaways with user
- Create a summary page in
wiki/sources/ - Create or update entity pages in
wiki/entities/ - Create or update concept pages in
wiki/concepts/ - Update
wiki/index.md - Append entry to
wiki/log.md
Query
When user asks a question:
- Read
wiki/index.mdto find relevant pages - Read relevant wiki pages
- Synthesize answer with citations to wiki pages
- If answer is valuable, offer to save as
wiki/synthesis/page
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 72 lines · 621 tokens per session scan A d874abea0d5e
memoriki CLAUDE.md is an instructions file published in the GitHub repository AyanbekDos/memoriki (120 stars, last pushed 4mo ago), licensed MIT. It adds 621 tokens to every session, about $0.0031 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
lorekit CLAUDE.md
Instructions for GYF0311/lorekit, a project described as: Personal LLM Wiki toolkit — let AI build and maintain your knowledge base. Based on Karpathy's LLM Wiki pattern. TypeScript, ollama+bge-m3, works with any AI coding agent.
llm-wiki-agent CLAUDE.md
Claude Code instructions for SamurAIGPT/llm-wiki-agent, covering llm wiki agent — schema & workflow instructions, slash commands (claude code), directory layout, page format and ingest workflow.
pi-llm-wiki AGENTS.md
AGENTS.md instructions for zosmaai/pi-llm-wiki, covering agents.md, project, tech stack, file layout and conventions.
mindbase-llm-wiki CLAUDE.md
Claude Code instructions for frankchu91/mindbase-llm-wiki, covering mindbase — north star for claude code, product thesis (one paragraph), the 3-layer architecture (from karpathy), the 3 operations (from karpathy) and special files (from karpathy).
hippocampus CLAUDE.md
Instructions for sturlese/hippocampus, covering layout, frontmatter schema, conventions, answering questions from the vault (read order) and contradictions.
llm-wiki AGENTS.md
Instructions for cobusgreyling/llm-wiki, covering llm wiki agent instructions, architecture, page types, special files and log entry format.