Skip to content

po4yka/llm-wiki-skills / skills/llm-wiki-orient

llm-wiki-orient

Verified

Explain the LLM-Wiki pattern, its raw/wiki/schema architecture, how it differs from RAG, GraphRAG and agent memory, and what ready-made solution families exist. Use when the user asks what LLM-Wiki is, whether it is worth adopting, what OpenWiki/wiki memory means, or how the ecosystem fits together.

LLM-Wiki Orient

Goal

Give the user a clear, non-hype explanation of LLM-Wiki and route them to the next useful skill.

When to use

Use when the user asks:

  • "What is LLM-Wiki?"
  • "How is this different from RAG?"
  • "What tools exist?"
  • "Is this useful for my notes/team/repo?"
  • "Explain OpenWiki/wiki memory/compiled knowledge."

For objections such as "why do I need this?", "what proof exists?", "what benefits will I get?", "how do I keep it alive?", or "do I need Obsidian?", route to or invoke llm-wiki-faq.

Inputs

  • The user's question about LLM-Wiki (what it is, how it differs from RAG/GraphRAG, whether to adopt it).
  • Optional context frame: personal second brain, team/company knowledge, repository documentation, research corpus, or product build.
  • Optional local-first requirement and sensitivity/privacy level.
  • Optional target agent: Claude Code, Codex, Cursor, OpenCode or other.
  • Whether the user already has documents to migrate or is starting empty.

Procedure

1. Identify the user's frame

Infer or ask for the minimum needed context:

  • personal second brain, team/company knowledge, repository documentation, research corpus, or product build;
  • local-first requirement;
  • target agent: Claude Code, Codex, Cursor, OpenCode or other;
  • existing documents or empty start;
  • sensitivity/privacy level.

If the user only wants a general explanation, do not interrogate them.

2. Explain the pattern

Use this mental model:

raw/     immutable source material
wiki/    LLM-compiled, human-readable Markdown knowledge layer
schema/  AGENTS.md, CLAUDE.md, skills, schemas and conventions

Emphasize:

  • LLM-Wiki is a compiled knowledge layer, not just a note app.
  • Raw sources stay preserved.
  • The wiki is reviewable Markdown.
  • Skills define procedures over the wiki.
  • index.md and log.md make the wiki navigable and auditable.
  • Evidence and adoption questions should be answered with direct/adjacent/local evidence levels, not hype.

3. Compare to neighboring patterns

Explain briefly:

PatternDifference
RAGRetrieves chunks at query time. LLM-Wiki pre-compiles durable pages and still may use search.
GraphRAGBuilds graph-aware retrieval structures. LLM-Wiki makes human-readable Markdown the primary artifact.
Agent memoryOften black-box or agent-owned. LLM-Wiki is inspectable, git-versioned and user-owned.
CLAUDE.md / AGENTS.mdSchema/instruction layer, not the whole knowledge base.
SkillsOperators that perform tasks over the wiki.

4. Present solution families

Use current browsing for volatile facts. The stable categories are:

  • repo-documentation agents such as OpenWiki-style tools;
  • Obsidian/local-first Markdown workflows;
  • hybrid search add-ons such as qmd-style local retrieval;
  • graph/RAG systems when machine retrieval is primary;
  • custom CLI/plugin/product builds;
  • team PR-based wiki maintenance workflows.

5. State the trust boundary

Say clearly:

Automate bookkeeping, not belief.

The agent can maintain links, indexes, summaries, metadata, triage reports and lint reports. Human review should own final synthesis, ambiguous claims and promotion to verified knowledge.

6. Route to next skill

Recommend one next skill:

  • answer adoption objections or evidence questions: llm-wiki-faq;
  • choose a solution: llm-wiki-choose;
  • install/configure: llm-wiki-setup;
  • migrate existing docs: llm-wiki-refactor;
  • build custom: llm-wiki-design;
  • get current news: llm-wiki-news-radar;
  • operate an existing vault: wiki-triage, wiki-ingest, wiki-query, wiki-lint.

Output

Use this shape:

## What LLM-Wiki is

## What it is not

## Main solution families

## Evidence level

## Fit for your case

## Recommended next step

Safety gates

  • Do not present generated summaries as verified truth.
  • Do not recommend a specific external tool using stale maturity facts.
  • Browse for current project status, releases, pricing, compatibility or maintenance claims.
  • Do not claim LLM-Wiki is proven universally better than RAG; route evidence questions through llm-wiki-faq.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall