Skip to content

Index

Catalog of all wiki pages. One line each: link + summary. Updated on every ingest. Read this first when answering questions from the wiki.

Topics

Grouped by domain folder under topics/. A page has exactly one physical home; its cross-cutting axes live in the frontmatter tags: and in each page's ## Related section.

AI & Knowledge Systems (ai/)

  • ai-agent-architecture — the agent loop, memory, context engineering, MCP, sandboxes, cost architecture. #ai #guide
  • claude-code-memory-architecture — Claude Code auto memory (injected schema, MEMORY.md index, no dedicated tools, recall with no retrieval engine) vs the API memory_20250818 tool; schema by convention vs enforcement. #ai #pkm #guide
  • agent-memory-at-scale — when file-and-index memory stops scaling (the index breaks, not the storage), the three real triggers for a vector DB, and RAG as an agent search tool. Flags a tension with ai-agent-architecture §6.4. #ai #pkm #guide
  • ai-adoption-maturity-model — the five steps of org AI adoption (0 gated → 4 AI-native): the bottleneck that defines each step, the transition unlocks, and how guardrails escalate in kind. Flags a tension with ai-agent-architecture §2.3 on multi-agent. #ai #devops #guide
  • llm-wiki-pattern — Karpathy's compile-don't-retrieve pattern: three layers (immutable sources / LLM-owned wiki / the schema), three operations, index vs. log. The pattern this vault implements — includes where BRAIN.md diverges from it. #ai #pkm #guide
  • okf-spec — Google Cloud's markdown + YAML-frontmatter spec for agent-consumable knowledge: one required field, bundles as directory trees, interoperability surface vs. content model. Tested against agent-memory-at-scale's index-scaling argument. #ai #pkm #data #guide
  • llms-txt-vs-robots-txt-vs-sitemap — three machine-readable files compared; the publishing side of the context problem ai-agent-architecture §7 describes from the agent's end. #web #ai #guide

.NET (dotnet/)

Security & Identity (security/)

Ops & Infrastructure (ops/)

Design & UX (design/)

  • ux-psychology-principles — why interfaces work: perception, memory, interaction cost, decisions, motivation, trust, each with its mechanism and an evidence grade (replicated / contested / folklore), plus corrections to famous claims. #design #ux #psychology #guide
  • dark-patterns — persuasion turned against the user: Brignull → Gray → Mathur taxonomies, DSA Art. 25 and FTC anchors, and the three-check test (truthful framing, symmetric exit, endorsement). #design #ux #ethics #regulation #guide
  • ui-craft-rules — grid, palette, type, icons, dividers, CTAs, each tied to the mechanism it serves; WCAG thresholds as the same mechanisms at their limit. #design #ux #accessibility #guide

Entities

(none yet — see the growth frontier below)

Sources

Provenance of the wiki pages above (updated 2026-07-20). The first bucket is exactly the set of pages carrying a sources: frontmatter field, so it is mechanically checkable: - Draws on a source in sources/: agent-memory-at-scale, ai-adoption-maturity-model, dark-patterns, llm-wiki-pattern, okf-spec, ui-craft-rules, ux-psychology-principles. (agent-memory-at-scale began as a session synthesis and later took substantive content from two sources — the sources: field records what fed a page, not where it started.) - Filed back from a working session, no raw source: claude-code-memory-architecture. - All others were authored by Wim in the obsidian-notes vault and migrated 2026-07-19.

Growth frontier

Wikilinks that exist in the pages above but have no page yet — each marks something worth writing, in the llm-wiki-pattern sense. All originate in ai-agent-architecture unless noted.

Entity candidates: [[LangChain]] · [[Model Context Protocol]] · [[Microsoft Agent Framework]] · [[OpenTelemetry]] · [[Google Cloud]] (from okf-spec — deliberately not created 2026-07-20; one blog post is thin grounds to open the entities section)

Also declined 2026-07-20: an [[Andrej Karpathy]] entity page, same reasoning as [[Google Cloud]] — one gist is thin grounds. The entities section stays empty by decision, not by neglect; revisit when a second source makes the same actor recur.

Topic candidates: [[ReAct]] · [[prompt injection]] · [[JSON Schema]] (also in xml-transformation-framework)

Retired 2026-07-19: [[RAG]] and [[vector database]] — the useful knowledge (when either is warranted, and how they fit an agent loop) is now written up in agent-memory-at-scale. Dedicated stubs would only be generic explainers. The bare wikilinks still appear in ai-agent-architecture and stay unresolved by design; its Related-notes list points at the covering page instead.

Closed 2026-09-06. The four clusters carried over from the old vault's Home.md are now written as ## Related sections on the pages themselves, in both directions: [[csharp-xsd-schema-tools]] ↔ [[xml-transformation-framework]]; the identity stack ([[oauth-oidc-keycloak-architecture-guide]] ↔ [[dotnet-microservices-auth-architecture]] ↔ [[seven-cryptography-concepts]]); [[ai-agent-architecture]] ↔ [[llms-txt-vs-robots-txt-vs-sitemap]]; and the reciprocal on [[postgresql-backup-restore-docker]] ↔ [[odoo-filestore-backup-restore-docker]], which had only been one-way.

The same pass reached the nine pages migrated on 2026-07-19 that had no inbound link from any page — their only reference had been their catalog line above. Every topic page now has at least one ## Related edge.