Candidate Obsidian Affordances
This is an initial inventory, not a recommendation. The goal is to separate "useful interface adaptation" from "representation drift."
Likely ready to explore
- Readable frontmatter/property ergonomics. Commonplace already uses frontmatter heavily. Check whether small naming or formatting choices make the repo more comfortable in Obsidian without changing semantics.
- Backlink and local-graph friendliness. The repo already has strong links; the question is whether small compatibility tweaks improve Obsidian's existing graph/backlink surfaces without changing canonical link syntax.
- Vault-local configuration. A minimal
.obsidian/config could improve reading experience, saved searches, or panes without changing the knowledge substrate itself. - Template support for workshop scaffolds. Obsidian templates could make workshop setup faster for humans if they remain optional and do not become a second canonical write path.
- Obsidian Web Clipper as the source-capture front-end. This is the strongest concrete Obsidian affordance identified so far. It looks promising as a replacement for
/snapshot-webin browser-driven capture workflows while preserving/ingestas the graph-aware analysis and recommendation layer.
Interesting but needs a concrete use case
- Bases-style query views. Structured slices over frontmatter and workshop state may be genuinely useful, but only if there are recurring questions that grep and indexes answer poorly.
- Canvas views for active workshops. This could help design or triage work, but only if a workshop genuinely benefits from spatial arrangement rather than prose notes and links.
- Bookmarks and workspace layouts. These are useful for a live operator, but may be too user-specific to commit unless they encode a shared operational workflow.
- Aliases and display names. Helpful in Obsidian, but risky if they blur the KB's title-as-claim and filename discipline.
- Wiki-link syntax or dual-syntax support. Potentially valuable for Obsidian ergonomics, but only if explicit relationship semantics, migration cost, tooling updates, and canonical-syntax rules are all handled deliberately.
Probably reject or keep non-canonical
- Semantically thin graph usage as a substitute for relationship language. Useful as a view, not as a replacement for articulated link meaning.
- Obsidian-only metadata that becomes required for canonical notes. If a field only exists to satisfy one tool and adds no retrieval or reasoning value, it should stay optional or generated.
Evaluation criteria
- Does the affordance improve human navigation or agent navigation in a way we can describe concretely?
- Can it stay optional or generated?
- Does it preserve standard markdown links and explicit relationship semantics?
- Does it preserve clean git diffs and repo reviewability?
- Does it help the workshop layer, the library layer, or both?
- Would a simpler alternative in plain markdown or shell scripts achieve the same result?