Meet K-Plex: The New Obsidian Graph View Plugin  🔗 **Source:** [Meet K-Plex: The New Obsidian Graph View Plugin](https://www.youtube.com/watch?v=pQbv5jSaW9Q) > [!note]+ Document Notes > tags: [obsidian-plugins, knowledge-graph, ontology, dataview, excalibrain, visual-pkm] > author: Zsolt Viczián > source: Zsolt's Visual Personal Knowledge Management (YouTube) > reference: https://www.youtube.com/watch?v=pQbv5jSaW9Q > date created: 2026-09-23 10:09 > > > [!info] Generated analysis — not the author's words. Sections below are machine-written > > from the captured source. Only text inside `> [!quote]` blocks is verbatim from the author. > > Do not quote or attribute any other line to the named author. > > # 🏷️ Title: Meet K-Plex: The New Obsidian Graph View Plugin (ExcaliBrain 2.0) > <!-- deep-read:sections start --> ## 🧭 Table of contents - [📝 Summary](#-summary) - [📖 Main content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning styles assessment](#-learning-styles-assessment) - [📊 Accessibility assessment](#-accessibility-assessment) - [🌍 DEIB assessment](#-deib-assessment) - [✅ Logical validity](#-logical-validity) - [🛡️ DARVO analysis](#️-darvo-analysis) - [🎯 Conclusion](#-conclusion) - [🔄 Refactor appendix](#-refactor-appendix--meet-k-plex-the-new-obsidian-graph-view-plugin-excalibrain-20) ## 📝 Summary - **What happened.** Zsolt Viczián, author of the Excalidraw and ExcaliBrain plugins, published a 12:13 walkthrough on 2026-09-22 of K-Plex (spoken as "Knowledgeplex"), a new Obsidian plugin he positions as the replacement for ExcaliBrain. The plugin had been in the Obsidian community directory for two days at recording time (00:11:32). The auto-transcript spells the name "K-Plex" everywhere except the install segment, where it renders "K-Plax" twice; quotes below preserve that spelling. - **The claim.** K-Plex is a structured knowledge graph: every note's relationships are sorted into parents, children, friends, and challengers by an ontology you define, and each type has a fixed position on screen (00:00:47). It reads relationships from both frontmatter properties and Dataview inline fields (00:01:33 to 00:02:20), which the speaker says Obsidian's built-in Bases feature does not. It imports existing ExcaliBrain settings on first run (00:06:56). - **The mechanism.** The plugin indexes the vault (a red dot signals indexing in progress, 00:04:35), builds typed edges from frontmatter and `field:: [[link]]` syntax, and renders the active note at the center with typed "gates" fanning out to related nodes. Right-clicking an edge explains which field in which note created it (00:02:20 to 00:03:05). Nodes can be expanded by markdown section, created directly from the graph with live search, pinned, filtered, and browsed by folder or tag (00:03:51 to 00:08:27). - **Why it matters.** Obsidian's default graph view is a force-directed hairball; it shows that notes connect but not how. A positional, ontology-driven graph turns the same links into a navigable map, and reading Dataview inline fields means vaults built on that convention (the reader's included) get typed edges without migrating metadata to frontmatter. - **Key stakeholders.** Zsolt Viczián as sole developer; ExcaliBrain users whose settings carry over; Dataview-heavy vault owners; the Sketch Your Mind community, which the video names as the feedback channel; Obsidian itself, whose Bases and Graph View the speaker criticizes directly. - **Risks and tradeoffs.** The plugin is self-described as "heavily in development" (00:10:45) at version 0.0.4 per the description link. The speaker gets lost in his own settings menu at 00:08:27 to 00:09:15, a live signal that the UI is not yet discoverable. Deep child expansion becomes "very busy" on large graphs by his own account (00:09:15). Folder and tag indexing adds visible latency. Every feature claim here is a demo on the developer's own vault, not an independent test. - **Bottom line.** For anyone already running ExcaliBrain or storing relationships in Dataview inline fields, K-Plex is the upgrade path and costs nothing to try. The strongest idea in the video is the design principle, not the plugin: a graph is useful when position carries meaning and every edge can explain itself. ## 📖 Main content **The design argument.** The video opens by borrowing Simon Sinek's "start with why" (00:00:00) and answers it at 00:00:47: this is a structured graph, meaning each idea has a fixed place on screen determined by its relationship type. Parents, children, friends, and challengers each occupy a region; the current note sits at the center; navigating to a neighbor recenters the graph and the accompanying editor pane follows (00:00:47). The transcript states explicitly that challenger relationships render on the right-hand side (00:05:22). The other positions are not spoken aloud; they follow ExcaliBrain's convention of parents above, children below, friends to the left, and viewers should confirm them in the plugin rather than from this analysis. **Ontology as the source of truth.** Every relationship the graph draws comes from an ontology the user defines (00:01:33): a mapping from field names (the demo shows `source`) to relationship types. The plugin reads those fields from two places. First, Obsidian document properties (frontmatter). Second, Dataview-style inline fields written as `field:: value` in the note body. The speaker's stated reason for building this is that Obsidian's Bases feature reads only frontmatter and "does not query on the data view type fields" (00:02:20). That is a claim about Bases at the time of recording; Bases is built on the properties system, so the mechanism is plausible, but check the current Obsidian release notes before repeating it as fact. **Explainable edges.** At 00:02:20 the speaker right-clicks a relationship line and asks the plugin to explain it. The explanation names the field and note that produced the edge, and offers to jump to each occurrence when the field appears in more than one place (00:03:05). This is the feature that most separates a structured graph from a force-directed one: an edge in the default Obsidian graph has no provenance, while an edge here can be traced to a line of markdown. **Gates, sections, and node creation.** Hovering a node shows its outgoing links; hovering a "gate" (the connector for one relationship type) shows every node reachable through it (00:03:05). A node can be expanded into its markdown sections so the graph shows which section a given link originates from (00:03:51). Dragging from a gate creates a new note of that relationship type; a search box fills as you type, the ontology field can be switched before committing, and Ctrl+Enter creates the note (00:03:51 to 00:04:35). The red indexing indicator blinks while the vault re-indexes after the change (00:04:35). Adding a `challenges` field to a note places the target on the challengers side immediately (00:05:22). **Layout, pins, siblings, filters.** The graph can be moved below the editor, hidden temporarily, or pinned to a specific note so it stops following navigation (00:05:22). Line style is configurable. Siblings, which the speaker defines as "more distant thoughts" (00:06:08), can be toggled on; the demo note *Finite and Infinite Games* (the transcript renders it "the finite and infinite game," James Carse's 1986 book) shows a long sibling list, which a filter box narrows (00:06:08). The filter term in the transcript is "card items," which is likely a caption error; the intended term is unclear from audio alone. **Settings and ExcaliBrain migration.** Settings cover the graph, the ontology, appearance, and the sidecar pane (00:06:08 to 00:06:56). On first run the plugin loads existing ExcaliBrain data, so an ExcaliBrain user's ontology carries over (00:06:56). **Folders and tags as graph nodes.** The speaker's sharpest criticism lands at 00:06:56: he loves Obsidian, but calls the built-in Graph View "a big mistake." His alternative is to add folders and tags to the graph so the vault's physical structure and its tag taxonomy become navigable trees alongside note links (00:06:56 to 00:08:27). Indexing that data takes visible time. The demo walks a folder hierarchy to a storyboard-building note, then navigates a `book/fiction` tag to list every fiction book he has read (00:07:41 to 00:08:27). Relationship labels on lines can be hidden in settings (00:08:27). **Deep hierarchy view.** After a detour through the menu ("it will be amazing if I remembered how to do that," 00:08:27), the speaker finds "expand to view," which shows children of children for a multi-level look (00:09:15). He concedes it gets busy on a large graph and says he prefers the compact default, with detail and height adjustable for larger screens (00:09:15 to 00:10:01). **Sidecar and mobile.** The sidecar pane opens a note beside the graph and can dock left or right (00:10:01). A mobile version opens the graph from the side panel (00:10:45). **Status and call to action.** The plugin is new, in heavy development, and the speaker wants feature ideas on the Sketch Your Mind community forum, where he says some requests are already being implemented (00:10:45 to 00:11:32). Install path: Settings, Community plugins, Browse, search for K-Plex (00:11:32). **What the video does not establish.** No performance numbers, vault sizes, or timings are given. "Tons more than ExcaliBrain" (00:00:00) and "much, much, much more than ExcaliBrain ever was" (00:10:45) are not itemized against a feature list. The Bases limitation is asserted, not demonstrated. The graph's behavior on vaults of thousands of notes is described only as "very busy" for the deep view. ## 🗒️ Notes #### ⏱️ [00:00:00](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=0) Introducing K-Plex (Knowledgeplex) - Speaker names himself and the plugin, states it replaces ExcaliBrain, and claims it already exceeds ExcaliBrain's feature set with more planned. - Frames the pitch with Simon Sinek's "start with why" and promises to explain how the graph differs from others. > [!quote] [00:00:00] > So this is going to be your graph database, your graph UI for Obsidian on steroids. #### ⏱️ [00:00:40](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=40) Why K-Plex? Structured knowledge graphs and ontologies - Four relationship regions named: friends, challengers, parents, children. The central node is the note being viewed. - Navigation demo: Simon Sinek, then his book, then the "start with why" illustration. The editor pane follows the graph. - All node placement follows a user-supplied ontology. > [!quote] [00:00:47] > This graph is a structured knowledge graph, which means that ideas have their specific places on the graph. #### ⏱️ [00:01:44](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=104) Dataview inline fields and frontmatter properties support - Test fixture "node A" exercises every relationship type; the speaker built it while developing the plugin. - Relationships come partly from Obsidian document properties. The speaker's stated reason for not using Bases: it reads only properties. - Dataview inline fields (the demo shows `source`) also define relationships. The transcript renders "Dataview" as "data view." > [!quote] [00:01:33] > one reason I don't really use bases is because it only uses the document properties. > [!quote] [00:02:20] > It does not query on the data view type fields. #### ⏱️ [00:02:45](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=165) Inspecting and explaining relationships - Right-click on an edge, choose "explain," and the plugin shows which field produced it and why it sits where it does. - When a field appears in several places, the explanation lists each and lets you jump to it. > [!quote] [00:02:20] > I can right click on this and I can ask this to be explained #### ⏱️ [00:03:20](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=200) Gate connections and expanding note sections - Hover a node to see where it points; hover a gate to see every node reachable through that relationship type. - Right-click menu on a node includes "expand into sections." Density setting increased on screen for legibility. - Expanded nodes show links grouped by the markdown section they come from; sections collapse individually. > [!quote] [00:03:05] > If I point to a gate, then the gate is going to show me all the nodes that go from that gate. #### ⏱️ [00:04:06](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=246) Creating nodes directly from the graph interface - Drag from a gate to create a child; live search shows matching existing notes as you type. - Ontology field can be changed before committing; Ctrl+Enter creates the note. - Red dot in the toolbar blinks while re-indexing after the change. - Adding a `challenges` field to the new note places the target on the challengers side (right) at 00:05:22. > [!quote] [00:03:51] > you can also add new child, parents or whatever. > [!quote] [00:04:35] > I had the red dot blinking because that means that indexing when there's a change indexing is ongoing. #### ⏱️ [00:05:32](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=332) Pinning nodes, layout controls and sibling filters - Layout: move the graph below the editor, hide it temporarily, hide it while writing and bring it back. - Pin the graph to any note so it stops following navigation. Line style is configurable. - Siblings are "more distant thoughts"; *Finite and Infinite Games* has many, and a filter box narrows the list (transcript renders the filter term as "card items," likely a caption error). > [!quote] [00:05:22] > you can actually pin the graph to any node in your Obsidian Vault > [!quote] [00:06:08] > You can add the siblings and siblings are more distant thoughts. #### ⏱️ [00:06:44](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=404) Graph settings and ExcaliBrain compatibility - Settings sections: graph, ontology, appearance, sidecar. - First run imports existing ExcaliBrain data. - Direct criticism of Obsidian's built-in Graph View, paired with a statement of affection for Obsidian itself. > [!quote] [00:06:56] > all of this is ExcaliBrain compatible. So the first time you run this, it is going to load the ExcaliBrain data in here. > [!quote] [00:06:56] > I love Obsidian, but the GraphView is in my view, a big mistake. This is the type of graph we would need. #### ⏱️ [00:07:15](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=435) Navigating folders and tags in the graph - Folders and tags can be added as graph nodes; indexing pulls the extra data with visible delay. - Folder walk demo ends at a storyboard-building book note. Tag walk from *Finite and Infinite Games* to `book/fiction` lists every fiction book read. - Sibling filter works here too; relationship labels on lines can be hidden in settings. > [!quote] [00:07:41] > this will give you the opportunity to explore your notes. > [!quote] [00:08:27] > If you don't like these tags on the lines, which explain the relationship between the book and the child, you can hide that in settings. #### ⏱️ [00:08:52](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=532) Expanded deep child hierarchy view - Speaker hunts through the menu for the option, passing "show/hide inferred relationships" and pin sets before finding "expand to view." - Expanded view shows children of children. Speaker concedes it gets busy on a large graph and prefers the compact default. - Detail level and node height are adjustable for larger screens. > [!quote] [00:09:15] > Now, of course, with a big graph, this becomes very busy. #### ⏱️ [00:10:10](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=610) Graph walk-through, sidecar view and mobile experience - The stated core benefit: walking around your notes rather than searching them. - Sidecar pane opens a note beside the graph; speaker prefers it docked right and narrows the graph to fit. - Mobile version opens the graph from the side panel. > [!quote] [00:10:01] > you can start to experience your notes as they should be experienced. > [!quote] [00:10:45] > this also has a mobile version, then the graph opens from the side panel. #### ⏱️ [00:11:15](https://www.youtube.com/watch?v=pQbv5jSaW9Q&t=675) How to install K-Plex and join the discussion - Status: new, heavily in development, feedback wanted on the Sketch Your Mind community site (linked in description). Some requested features already being built. - Install: Settings, Community plugins, Browse, search "K-Plex." Transcript renders the name "K-Plax" here. Listed two days before recording, few downloads. > [!quote] [00:10:45] > It's, at this time, still heavily in development. > [!quote] [00:11:32] > right now it's brand new two days ago. ## 📚 References The transcript itself contains no links; the following come from the video description or are primary sources for named entities. - [Meet K-Plex: The New Obsidian Graph View Plugin (ExcaliBrain 2.0)](https://www.youtube.com/watch?v=pQbv5jSaW9Q): the source video, 12:13, uploaded 2026-09-22. - [K-Plex v0.0.4, a major leap (Sketch Your Mind community)](https://community.sketch-your-mind.com/t/k-plex-v0-0-4-a-major-leap/11153): the discussion thread the speaker points to at 00:10:45; first link in the description. - [K-Plex on GitHub](https://github.com/zsviczian/kplex): source repository; second link in the description. - [ExcaliBrain on GitHub](https://github.com/zsviczian/excalibrain): the predecessor plugin whose data K-Plex imports (00:06:56). - [Obsidian Excalidraw plugin](https://github.com/zsviczian/obsidian-excalidraw-plugin): the speaker's other plugin, linked in the description. - [Obsidian Bases](https://help.obsidian.md/bases): the built-in feature the speaker says reads only frontmatter properties (00:01:33). Primary source for checking that claim against the current release. - [Dataview plugin](https://blacksmithgu.github.io/obsidian-dataview/): origin of the `field:: value` inline-field syntax K-Plex reads (00:02:20). - [Obsidian](https://obsidian.md): host application; community plugin directory is the install path (00:11:32). - *Start with Why*, Simon Sinek (Portfolio, 2009): the framing device at 00:00:00 and the demo notes at 00:00:47. - *Finite and Infinite Games*, James P. Carse (Free Press, 1986): the demo note the transcript renders as "the finite and infinite game" (00:06:08, 00:07:41, 00:08:27). - [Sketch Your Mind community](https://community.sketch-your-mind.com/welcome): the speaker's forum, named at 00:10:45 as the feedback channel. ## 🗣️ Paraphrase Zsolt Viczián has replaced ExcaliBrain with a plugin he calls K-Plex, and his argument for it starts with a complaint about how graphs usually work. A force-directed graph tells you that two notes connect and nothing more. His graph puts every neighbor in a place that means something: parents in one region, children in another, friends and challengers on their own sides, all decided by an ontology you write yourself. Click a neighbor and the whole graph recenters while the note opens beside it. Right-click an edge and the plugin tells you which field in which note created it. The second half of his case is about where relationships live. Obsidian's Bases reads frontmatter properties only. Many vaults, his included, carry relationships as Dataview inline fields in the note body. K-Plex reads both, so a `source:: [[Note]]` line becomes a typed edge without moving anything into YAML. From there the video is a tour: expand a node by section, drag a gate to create a typed child with live search, pin the graph, filter siblings, browse folders and tags as if they were notes, expand children of children until it gets too busy, dock a sidecar, and open it all on mobile. He closes by admitting the plugin is two days old and asking for feature requests on his community forum. ## 📰 Newsletter **Your graph view should know what an edge means.** Zsolt Viczián, the developer behind Excalidraw for Obsidian, has retired ExcaliBrain and shipped its successor: [K-Plex](https://github.com/zsviczian/kplex), a structured graph plugin where parents, children, friends, and challengers each get a fixed spot on screen and every line can explain which field created it. The [12-minute walkthrough](https://www.youtube.com/watch?v=pQbv5jSaW9Q) went up 2026-09-22, two days after the plugin landed in the community directory. Why it matters for the WordPress-adjacent PKM crowd: if you document plugin architecture, release dependencies, or handbook lineage in Obsidian, you have probably been writing `depends-on::` or `supersedes::` as Dataview inline fields for years. Obsidian's Bases can't see those. K-Plex can, and it turns them into typed edges you can walk. It also imports your ExcaliBrain settings on first run, so the migration cost is one install. The honest caveats from the video itself: it's version 0.0.4, the developer lost his own settings menu on camera, and the deep-hierarchy view gets crowded fast on a big vault. One thing the video doesn't mention: this is a client-side plugin, so nothing it renders reaches Obsidian Publish or any static export. Your published site still gets the plain links. Feature requests go to the [Sketch Your Mind thread](https://community.sketch-your-mind.com/t/k-plex-v0-0-4-a-major-leap/11153). ## ✅ Action items #### DevRel - Developer advocacy: none for Pantheon products. Nearest relevance: the "explain this edge" pattern (00:02:20) is a good model for any dependency-graph UI in developer tooling; cite it when reviewing graph features. - Education: a 20-minute internal show-and-tell on "graphs with position semantics vs force-directed graphs," using the K-Plex vs Obsidian Graph View contrast at 00:06:56 as the concrete case. - Developer experience: no direct action. Nearest relevance: K-Plex's live-search-on-create (00:03:51) is a reference point for autocomplete UX in scaffolding tools. - Developer marketing: none. This is a personal-vault tool. - Community management: none. Nearest relevance: the speaker's forum-first feedback loop (00:10:45) is a low-cost pattern for pre-1.0 tools. - Internal/external/cross-project: none from this source. #### OSPO - Policy and governance: none. Nearest relevance: a solo-maintainer plugin at v0.0.4 is the standard bus-factor case; if a team adopts it for shared documentation, note the single-maintainer dependency. - License compliance: confirm the K-Plex repository license before any org-wide vault template recommends it; the video does not name one. - Security management: community plugins run with full vault access. No action beyond the existing plugin-review norm. - Software inventory: if K-Plex enters a team vault template, record it alongside Excalidraw and Dataview as a dependency with a version pin. - Developer support: none. - Community engagement: the speaker accepts Ko-fi support (description). If the tool becomes a team dependency, a sponsorship is the proportionate ask. - Strategic alignment: none. #### Community - Maintainers: the video is a clean example of a maintainer replacing their own project and shipping a migration path (ExcaliBrain import at 00:06:56). Worth citing in any "how to deprecate" guidance. - Contributors: none directly. Nearest relevance: feature requests are being taken on a forum thread rather than GitHub issues (00:10:45), a lower-friction intake for non-developers. - Sponsors: none. - Ecosystem: none. #### WordPress - Core: none. Nearest relevance: dependency visualization for the plugin ecosystem is a recurring wish; K-Plex's positional layout is one answer to how such a view could read. - Core Performance: none. - Core AI: none. - **Core Program Team:** none. - Mobile: none. Nearest relevance: the graph opens from a side panel on mobile (00:10:45), a pattern for secondary views on small screens. - Playground: none. - Community: none. - Training: none directly. Nearest relevance: contributors who maintain learning-path notes in Obsidian gain typed prerequisite edges if they store `prerequisite::` as an inline field. - Marketing: none. - Support: none. - Design: none. Nearest relevance: right-click "explain" on a relationship (00:02:20) is a provenance affordance worth borrowing in any admin UI that shows derived data. - Documentation: contributors keeping handbook lineage in Obsidian can model page supersession as a `supersedes` relationship and browse it visually. No action for the team itself. - Photos: none. - TV: none. - Polyglots: none. - Accessibility: none from this source. Nearest relevance: the plugin's red-dot indexing indicator (00:04:35) is a color-only status signal; if anyone files upstream feedback, that is a concrete item. - Meta: none. - CLI: none. - Hosting: none. - Tide: none. - Openverse: none. - Themes: none. - Plugins: none. #### The WPOCC - Programs: none. Nearest relevance for The WP Open Community Collective: if fellows document governance relationships in Obsidian, typed edges (parent charter, child working group, challenger proposal) map directly onto K-Plex's ontology. - Projects: none from this source. - Funding experiments: none. ### FAIR - Technical independence: none directly. Nearest relevance: a local-first graph built from plain-text fields in the user's own files (00:01:33 to 00:02:20) is the independence property FAIR argues for; the data outlives the plugin. - Policies and covenants: none. - Productized governance artifacts: none. ## 🧪 Hypotheses 1. In vaults where at least 30% of typed relationships are stored as Dataview inline fields rather than frontmatter, K-Plex renders at least 30% more typed edges than a Bases view of the same notes, because Bases reads properties only (speaker's claim at 00:02:20). 2. Users locate a specific related note faster in a positional graph than in Obsidian's force-directed Graph View, with the gap widening as node degree rises, because relationship type is encoded in screen region (00:00:47). 3. ExcaliBrain vaults using default field names import into K-Plex with zero manual ontology remapping (00:06:56); vaults with renamed fields require manual fixes. 4. Indexing latency after enabling folder and tag nodes (00:07:15) grows with vault size and exceeds 5 seconds on vaults above 5,000 notes. 5. The deep child view (00:09:15) becomes unreadable above a fan-out threshold that can be measured, and the threshold sits below 30 visible nodes on a 1440px-wide display. ## 📊 Methodology - **H1.** Take one vault, count typed edges via a Dataview query over both frontmatter and inline fields, then count edges visible in a Bases table restricted to the same field names. Report the ratio. Repeat on three vaults with different frontmatter-to-inline mixes. - **H2.** Within-subject task study, 12 participants, 20 target-finding tasks each, counterbalanced between K-Plex and Graph View on the same vault. Measure time to first click on the correct node and error rate. Stratify by node degree (under 10, 10 to 30, over 30). - **H3.** Export five ExcaliBrain configurations (two default, three with renamed fields), install K-Plex fresh, diff the resulting ontology settings against the ExcaliBrain source. Count fields requiring manual edit. - **H4.** Generate synthetic vaults at 1k, 5k, 10k, 25k notes with fixed folder depth and tag counts. Time the indexing indicator from toggle to steady state, three runs each, on one fixed machine. Plot latency against note count. - **H5.** Fix display width at 1440px, use a synthetic tree with controllable fan-out, and ask participants to identify a labeled grandchild at fan-outs of 10, 20, 30, 50. Record accuracy and time. The threshold is the fan-out where accuracy drops below 90%. - **Controls throughout.** Same Obsidian version, same plugin version (pin the 0.0.x release), and record the ontology used, since the video shows results depend on it. ## ❓ Questions - Does the Bases limitation the speaker describes (00:02:20) still hold in the current Obsidian release, and does Obsidian plan inline-field support? - Which ExcaliBrain settings do not carry over on import (00:06:56)? The video says data loads but shows no diff. - How does K-Plex handle a field that appears in both frontmatter and as an inline field with conflicting values in the same note? - What is the indexing cost model? Is the vault re-indexed fully on every change, or incrementally? - Is the ontology stored in plugin settings only, or can it live as a note in the vault so it syncs and versions with the content? - Does the "explain relationship" feature (00:02:20) surface inferred relationships (the menu at 00:09:15 mentions "show hide inferred relationships") differently from explicit ones? - What does the mobile side-panel graph (00:10:45) omit relative to desktop? - Does a K-Plex graph export to anything static, given that Obsidian Publish and other exporters will never run the plugin? ## 🔮 Future research - **Ontology portability.** No shared standard exists for relationship-type vocabularies in personal vaults. A comparison of the ExcaliBrain/K-Plex default ontology against Juggl, Breadcrumbs, and the Idea Compass (parents, children, friends, competitors) would show how much overlap exists and whether a common set is feasible. - **Position semantics in graph UIs.** The claim that fixed regions per relationship type beat force layout (00:00:47) is testable and, as far as the video shows, untested. The H2 design above is the starting point. - **Inline fields vs properties as a long-term store.** The video treats reading both as a feature. The open question is whether vaults should keep relationships in the body at all, given that Obsidian's own tooling is converging on properties. - **Static rendering of structured graphs.** Publish pipelines drop plugin output. A build step that renders the K-Plex ontology to SVG per note would close the gap for published vaults and has no known implementation. - **Solo-maintainer plugin lifecycle.** ExcaliBrain to K-Plex is one data point in how single-developer tools get replaced rather than maintained. A survey across the Obsidian directory would show how common the pattern is and what migration paths look like. ## 💡 Implications - **Practice.** If your vault already uses the Idea Compass (parents, children, friends, competitors), K-Plex's four regions map onto it one-to-one; "challengers" is the competitors axis. The migration is an ontology mapping, not a note rewrite. Relationships stored as inline fields stay where they are. - **Policy.** Teams standardizing an Obsidian template should decide now whether relationships live in frontmatter, inline fields, or both. The video shows a tool that reads both; Obsidian's own Bases reads one. That asymmetry is a governance decision, not a tooling detail. - **Security.** A community plugin with vault-wide read access and an indexing loop is the normal Obsidian trust model. Nothing in the video changes that; nothing in it addresses it either. - **Education.** The video is a usable teaching case for "structure vs emergence" in graph design: the same links rendered two ways produce a hairball or a map. The 00:06:56 contrast is the slide. - **Funding models.** One developer, Ko-fi support, and a forum for feature requests. The tool's future depends on that continuing. Anyone building a workflow on it should either sponsor or accept the risk. - **Publishing.** For a vault that ships through Obsidian Publish, K-Plex improves the authoring experience and changes nothing for readers. Public structure still has to be built from links, Dataview output, or a static build step. ## 🧠 Learning styles assessment - **Reading level estimate.** Flesch-Kincaid Grade Level on the transcript body: 5.6 (1,741 words, 117 sentences, 2,275 syllables). Flesch Reading Ease: 81. The language is spoken and simple; comprehension depends on prior knowledge of Obsidian, Dataview, and ExcaliBrain, none of which the video defines. #### 👀 Visual learners - Present: the entire 12 minutes is a live screen demo of a visual tool; the graph, menus, and indexing indicator are all shown. - Improve: no static diagram of the four-region layout, so the one idea a viewer most needs (where each relationship type lands) is never shown as a labeled schematic. A single title card would fix it. The rapid cursor movement at 00:08:27 to 00:09:15 is hard to follow. #### 🔊 Aural learners - Present: continuous narration, no music bed competing with speech. - Improve: frequent deictic phrases ("click here," "this box right there," 00:03:05) carry no information by audio alone. Naming the menu item each time would make an audio-only pass usable. #### 🗣️ Verbal learners - Present: relationship types are named (parents, children, friends, challengers, siblings) and siblings get a one-line definition (00:06:08). - Improve: ontology, gate, sidecar, and inferred relationship go undefined. Bases and Dataview are assumed. Four sentences would cover them. #### 🧩 Kinesthetic learners - Present: install path is spoken step by step (00:11:32); the node-creation flow is shown end to end (00:03:51 to 00:05:22). - Improve: no sample vault or test fixture is shared, though "node A" (00:01:33) exists and would be the ideal download. GitHub link is in the description but never spoken. #### 🧠 Logical learners - Present: the video follows a why-then-how structure and states a mechanism for the Bases limitation (00:02:20). - Improve: no feature comparison table against ExcaliBrain or Graph View, so "tons more" (00:00:00) can't be checked. No performance numbers. #### 🧑‍🤝‍🧑 Social learners - Present: explicit invitation to the Sketch Your Mind forum with a note that requests are already shaping features (00:10:45). - Improve: the forum link is in the description only. No discussion prompt beyond "share ideas." #### 🧘‍♀️ Solitary learners - Present: short, self-contained, and chaptered; easy to rewatch a segment. - Improve: no reflection prompt. "Which of your relationships live in inline fields today?" would turn the video into an audit exercise. #### 🌿 Naturalistic learners - Present: the "walk around your notes" framing (00:10:01) is spatial and exploratory. - Improve: nothing ties the tool to systems or ecology thinking. Folder-and-tag trees as terrain would be an easy analogy to add. ## 📊 Accessibility assessment #### Screen reader semantics - Present: YouTube supplies page landmarks; the description carries a 12-entry chapter list with times, which reads well. - Improve: on-screen graph structure, menus, and the red indexing dot carry information the narration does not state. WCAG 2.2 AA 1.1.1 (Non-text Content) is unmet for the graph itself; a spoken description of the layout at 00:00:47 would cover most of it. #### Keyboard navigation - Present: the player is keyboard operable; chapters are reachable from the description. - Improve: nothing in the video adds barriers. 2.1.1 and 2.4.7 pass through the platform. #### Contrast and readability - Present: unverifiable from the transcript. The demo runs in the developer's own theme; the speaker increases density at 00:03:05 for legibility. - Improve: relationship labels on edges are small at default player size. Zooming during the explain-relationship demo (00:02:20) would help viewers who fail 1.4.3 (Contrast Minimum) thresholds on small text. #### Text resizing and reflow - Present: player scales; captions resize independently. - Improve: all UI text is burned into the video frame, so 1.4.4 (Resize Text) and 1.4.10 (Reflow) fail for in-frame text as they do for any screen recording. #### Media and captions - Present: an en-US caption track exists and is the basis of this analysis. Accuracy is high for common words; sentence boundaries are clean. - Improve: proper nouns fail. "K-Plax" for K-Plex at 00:11:32, "data view" for Dataview, "the finite and infinite game" for Carse's title, and "card items" for an unknown filter term. 1.2.2 (Captions, Prerecorded) is met in form and weak on the terms a viewer would search for. Correcting six terms fixes it. - Improve: no audio description. 1.2.5 (Audio Description, Prerecorded) is unmet, and the deictic narration style makes it matter more than usual. #### Forms and interactions - Present: the install flow is spoken step by step (00:11:32). - Improve: not applicable inside the video. Within the plugin, the red-dot indexing indicator (00:04:35) is a color-only status signal, which fails 1.4.1 (Use of Color) if it is the sole cue; upstream feedback item. #### Standards reference - Gaps against WCAG 2.2 AA: **1.1.1** (graph layout conveyed visually only), **1.2.5** (no audio description; deictic narration), **1.2.2** (captions accurate except proper nouns), **1.4.4 / 1.4.10** (burned-in UI text). Plugin-side: **1.4.1** (color-only indexing indicator). The chapter list is a strength and satisfies the spirit of 2.4.5 (Multiple Ways). ## 🌍 DEIB assessment #### Cultural breadth - Present: the speaker is a Hungarian developer presenting in English to a global audience; the tool is language-agnostic. - Improve: demo content is Anglo-American business and philosophy (Sinek, Carse). One non-English note in the demo would show the graph handles other scripts and would widen the reference set. #### Inclusive language - Present: "Hey everyone" opener; second-person throughout; no exclusionary terms. - Improve: "on steroids" (00:00:00) is a drug idiom that translates poorly. "Big mistake" (00:06:56) about Obsidian's Graph View is blunt but aimed at a feature, not a person. Low severity on both. #### Neurodiversity support - Present: 12 chapters with times in the description; one feature per segment; consistent demo pattern (show, click, explain). - Improve: the settings hunt at 00:08:27 to 00:09:15 is unedited wandering that costs a viewer with attention constraints 50 seconds. A cut or an on-screen "found it" marker would help. Undefined jargon adds load. #### Representation - Present: one presenter, self-identified, speaking about his own work. Credit to Simon Sinek and (by title) James Carse for the demo material. - Improve: no user voices. One clip of a community member's use case from the forum thread would show the tool outside the developer's vault. #### Psychological safety - Present: the speaker models getting lost in his own UI without apology or blame (00:08:27), invites disagreement via the forum, and says "don't get me wrong, I love Obsidian" before criticizing it (00:06:56). - Improve: "as they should be experienced" (00:10:01) frames one navigation style as correct. Viewers who prefer search or force layouts are told they're doing it wrong. "As I prefer to experience them" would remove that. ## ✅ Logical validity #### Strong reasoning - **Mechanism argument, 00:01:33 to 00:02:20.** Bases reads document properties; relationships in this vault also live in inline fields; therefore Bases misses those relationships. The premises are checkable and the conclusion follows. Clarify by naming the Obsidian version tested. - **Design argument, 00:00:47.** If each relationship type has a fixed screen region, then position carries information a force layout discards. That is valid as stated and is the strongest general idea in the video. Strengthen with a task-time comparison. - **Provenance argument, 00:02:20 to 00:03:05.** An edge that can explain which field produced it is auditable; an edge without provenance is not. Demonstrated live, with the multi-occurrence case handled. - **Concession against interest, 00:09:15.** The speaker says the deep view gets busy on large graphs and that he prefers the compact default. A limitation volunteered by the developer raises the credibility of the rest. - **Status disclosure, 00:10:45 and 00:11:32.** "Heavily in development," "brand new two days ago," and "not many downloads" are stated plainly. No inflation of maturity. #### Fallacies - **Bare assertion, 00:00:00 and 00:10:45.** "Doing tons more than ExcaliBrain ever did" and "much, much, much more than ExcaliBrain ever was" are unquantified. Weak because no feature list or count is given. *Fix:* a side-by-side feature table in the description. - **Opinion stated as verdict, 00:06:56.** "The GraphView is in my view, a big mistake" is hedged with "in my view" but "This is the type of graph we would need" is not. Weak because "need" is asserted for all users from one user's preference. *Fix:* name the tasks force layout fails at (finding a specific neighbor, tracing an edge) and let the claim rest on those. - **Normative framing, 00:10:01.** "Experience your notes as they should be experienced" assumes the conclusion that spatial navigation is the correct mode. *Fix:* "as I find most useful." - **Unverified comparison, 00:02:20.** The Bases limitation is stated, not shown. It may well be true; the video gives the viewer no way to check. *Fix:* a five-second Bases view of the same note showing the missing edge. - **No hidden interest.** The speaker built the plugin and says so at 00:00:00. This is advocacy for one's own work with the interest disclosed, so no suppressed-interest fallacy applies. Checked for appeal to authority (none; Sinek is a framing device, not a credential), false dilemma (none; other graph plugins are not mentioned at all), and slippery slope (none). #### Tone of voice - Enthusiastic, first-person, and unscripted. "Super excited" opens and closes the video (00:00:00, 00:11:32). Present tense and contractions throughout ("I can," "you can," "let's"). The register drops to self-deprecating during the menu hunt ("it will be amazing if I remembered how to do that," 00:08:27; "right now I'm a bit lost," 00:09:15) and returns to advocacy for the close. The pattern is a developer showing their own work with the rough edges left in, which reads as trustworthy and as unpolished in equal measure. ## 🛡️ DARVO analysis Not applicable to this source. DARVO (Deny, Attack, Reverse Victim and Offender) describes a response to accusation or criticism, and this video is a product walkthrough with no accuser and no defensive posture. For completeness across the three prongs: **Deny:** nothing is denied; limitations are volunteered (busy deep view at 00:09:15, heavy development at 00:10:45, few downloads at 00:11:32). **Attack:** the one critical statement, calling Obsidian's Graph View "a big mistake" (00:06:56), targets a feature, is prefaced with "I love Obsidian," and names no person. **Reverse victim/offender:** no such move appears. Emotional cues are enthusiasm and mild self-deprecation; no grievance rhetoric. ## 🎯 Conclusion K-Plex is ExcaliBrain's successor from the same developer, two days into public release at recording time, and the video shows enough working features to justify an install for anyone already running ExcaliBrain or storing relationships as Dataview inline fields. The ExcaliBrain import (00:06:56) makes the switch a one-step migration. The features that matter most are the ones the default Obsidian graph can't do: fixed screen regions per relationship type (00:00:47), edges that explain their own provenance (00:02:20), and typed node creation from the graph itself (00:03:51). The claims that don't hold up as stated are the comparative ones. "Tons more than ExcaliBrain" is never itemized, the Bases limitation is asserted rather than shown, and "as they should be experienced" turns one navigation preference into a rule. None of that is deceptive; the developer discloses his authorship, his plugin's immaturity, and its download count. It is a developer demoing his own work with the rough edges left in. **Next steps for readers:** install it, let it import the ExcaliBrain ontology, and check the Idea Compass mapping: parents, children, friends, and challengers line up with the vault's existing Parents / Children / Friends / Competitors convention. Then run one Dataview query counting relationships stored as inline fields versus frontmatter, because that ratio is how much K-Plex sees that Bases doesn't. And remember it changes nothing downstream: Obsidian Publish never runs the plugin, so published structure still comes from links and static output. ## 🔄 Refactor Appendix — Meet K-Plex: The New Obsidian Graph View Plugin (ExcaliBrain 2.0) > Density: 2–5 Literature atoms (H3 pad), 0–3 Permanent candidates (H4 pad). > Emitted: 3 Literature, 1 Permanent. Resonance filter applied — feature-tour claims > (pinning, sidecar, mobile) were dropped as implementation detail rather than ideas. > > Konik claim-header rules: present tense, declarative, ≤10 words, searchable, unique. > > Atomicity: one declarative sentence per atom. "and"/"also" → split. > > Tag pool: #bmf · #addendum · #xref · #articleseed · #storystem · #share · #aphorisms ### 📚 Literature Split Pad #### [H1] Structured graphs give each relationship type a place > [!quote] [00:00:47] > This graph is a structured knowledge graph, which means that ideas have their specific places on the graph. **Context & use**: - 00:00:47, the answer to the "start with why" framing; the design principle the whole plugin rests on. - Position encodes relationship type. That is what a force-directed layout throws away, and it maps one-to-one onto the Idea Compass (parents, children, friends, competitors) already used in this vault's visual summaries. **Tags**: #xref #addendum <!-- atomicity: 1 idea ✓ --> #### [H2] Bases reads properties only, not inline fields > [!quote] [00:01:33] > one reason I don't really use bases is because it only uses the document properties. **Context & use**: - 00:01:33, followed at 00:02:20 by "It does not query on the data view type fields." Speaker's claim, not demonstrated on screen. - Decides whether a vault's relationship metadata is visible to Obsidian's native tooling. Anyone with `field:: [[link]]` relationships should verify this against the current release before choosing where new metadata goes. **Tags**: #xref #share <!-- atomicity: 1 idea ✓ --> #### [H3] Force-directed graph view is the wrong default > [!quote] [00:06:56] > I love Obsidian, but the GraphView is in my view, a big mistake. This is the type of graph we would need. **Context & use**: - 00:06:56, immediately before adding folders and tags as navigable graph nodes. - An opinion, hedged then unhedged in one breath. Useful as the sharpest statement of the structure-vs-emergence argument, and as a specimen of a preference stated as a need. **Tags**: #xref #articleseed <!-- atomicity: 1 idea ✓ --> ### 🧠 Atomic Split Pad #### [P1] A graph edge is only as useful as its provenance **Claim** (one sentence, Courtney's voice): A link in a knowledge graph earns its place when I can trace it back to the line of text that created it. **Evidence pointer**: [[#H1]] · right-click "explain" demo at 00:02:20 to 00:03:05 **Scope & conditions**: Holds for graphs built from authored metadata (frontmatter, inline fields, typed links). Weakens for graphs built from inferred or statistical similarity, where provenance is a score rather than a source line. Breaks when the cost of authoring typed relationships exceeds the value of navigating them, which is the case for fast-capture notes that never get revisited. **Links**: - [[Idea Compass]] — because extends the four-direction navigation model into a rendered graph - [[Obsidian Graph View]] — because contrasts on whether position and edges carry meaning - [[Dataview inline fields]] — because applies-in the vault's existing metadata convention <!-- atomicity: 1 idea ✓ --> ### ✏️ Fleeting Queue - [F1] Verify the Bases claim at 00:02:20 against the current Obsidian release notes: does Bases read inline fields yet? — https://help.obsidian.md/bases #fleeting #inbox - [F2] Run a Dataview query counting relationships stored as inline fields versus frontmatter in the vault; that ratio is the size of what K-Plex sees that Bases doesn't #fleeting #inbox - [F3] Map the vault's Idea Compass (Parents / Children / Friends / Competitors) onto K-Plex's ontology; "challengers" should equal Competitors — confirm after import #fleeting #inbox - [F4] Open question: does anything render K-Plex structure statically for Obsidian Publish? If not, a per-note SVG build step is a gap worth an article #fleeting #inbox #articleseed - [F5] Transcript renders "card items" as the sibling filter term at 00:06:08; find the real term in the plugin UI before citing #fleeting #inbox - [F6] Article seed: "structure vs emergence in graph views," using the 00:06:56 contrast as the hook #fleeting #inbox #articleseed ### → Things 3 (max 2 actions) - [ ] Install K-Plex from Community plugins and confirm the ExcaliBrain ontology imports without remapping — vault maintenance, 20 min #things - [ ] Count vault relationships stored as Dataview inline fields versus frontmatter with one Dataview query — vault audit #things <!-- deep-read:sections end --> 📄 **Full text:** [[Full Document Contents/Articles/Meet K-Plex The New Obsidian Graph View Plugin.md|Meet K-Plex: The New Obsidian Graph View Plugin by Zsolt's Visual Personal Knowledge Management]] ## Highlights - [00:00:00] Hey everyone, it's Zsolt here. I am super excited to show you K-Plex, Knowledgeplex, a new Obsidian plugin I developed recently. This plugin is taking the place of ExcaliBrain and this plugin is already doing tons more than ExcaliBrain ever did and I have even more plans for it. So this is going to be your graph database, your graph UI for Obsidian on steroids. So let me show you how this works and let's start with why as Simon Sinek also recommends. So why this graph and how is it different from other graphs? [00:00:47] This graph is a structured knowledge graph, which means that ideas have their specific places on the graph. So here you can see the so-called friends, here you can see the challengers, here you can see the parents and you can see the children. If you want, you can open the central node. So that is actually the node we're looking at and if we maybe come back to Simon Sinek and then we come to his book and then we come to this illustration, start with why. As I'm navigating the graph, of course this side will navigate with the graph and of course at all times all of these nodes are going ([View Highlight](https://read.readwise.io/read/01m37a1zs2q6nqw2k2394ahc13)) - Note: [H1] Structured graphs give each relationship type a place - Tags: atom/literature/h1 - [00:01:33] to be structured based on an ontology that you provide. Now, compared to ExcaliBrain, you have some amazing new features. So first of all, let's look for node A. node A is part of a structure that I built to test all the relationships and to build the tool itself. And this also includes all of the different type of relationships. So first of all, you can see that this is based partly on the Obsidian document properties and that's great, but one thing I didn't like about bases and one reason I don't really use bases is because it only uses the document properties. [00:02:20] It does not query on the data view type fields. For example, you can see here source, source URL for line ontology and that source actually defines this relationship and indeed on this relationship, so this is effectively a URL here, this relationship, if I click on the relationship, I can right click on this and I can ask this to be explained and here I can understand why this is displayed here at the top and if I want, I can navigate here to this relationship that is going to show me where this link is coming from. ([View Highlight](https://read.readwise.io/read/01m37a21zjdmn2z268t104y5xp)) - Note: [H2] Bases reads properties only, not inline fields - Tags: atom/literature/h2 - [00:06:08] You can add, this is interesting. You can add the siblings and siblings are more distant thoughts. Let me maybe go back to the finite and infinite game. So here you can see that this node has lots of siblings. And by the way, you can see here that this has really many, many siblings here. But now you also have a filter here and I can say I want to see only the card items. And so then I can see a filtered list of all the siblings. There's plenty of settings that you can explore. So if you click here, you are taken to settings and you can set things around the graph. You can set things around the ontology, the [00:06:56] appearance, the sidecar functionality, etc. And by the way, all of this is ExcaliBrain compatible. So the first time you run this, it is going to load the ExcaliBrain data in here. And one thing again, I don't like about the Obsidian graph. There are many things I don't like about Obsidian graph and don't get me wrong. I love Obsidian, but the GraphView is in my view, a big mistake. This is the type of graph we would need. Here I can add folders to my GraphView and I can also add tags. And now of course, indexing is pulling in ([View Highlight](https://read.readwise.io/read/01m37a244f7ye4t61wtrxxey99)) - Note: [H3] Force-directed graph view is the wrong default - Tags: atom/literature/h3