You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Allow links outside a space to validate without warnings #128
Content links that point outside a space are reported as unresolved-link warnings, even when they're well formed and resolve elsewhere, e.g. in the wider Obsidian vault. sctx validate labels them "may be outside-space or broken", but they count as warnings all the same.
Measured across the eight spaces under one vault: 628 of 635 issues are unresolved content links (336 note names, 254 paths, 38 attachments such as .png and .base). One knowledge-wiki space has 427 of them against 4 errors.
This drowns out real issues in validate output, and it's why the plugin hooks' baselines are mostly content-link keys (#127).
Proposal
Let schemas or spaces declare which link fields accept values from outside the space as valid when they're otherwise well formed, instead of reporting a warning. Content links should still belong in the hooks: a link Claude introduces that doesn't resolve anywhere should still be reported.
Open questions
Where it's declared: schema (per link field, or for body content links), space config, or both.
How "well formed" is judged for links outside a space: syntax only, or resolving against a wider root such as the vault?
Whether attachments (.png, .pdf, .base, .canvas) need their own handling.
Another case worth deciding here, from a recorded agent session: a link whose target exists in the space but is excluded during parsing because the note has no type is reported under "Unresolved content links (may be outside-space or broken)" alongside genuinely missing targets. In a small test space, [[Untyped target]] (an existing, untyped note) and [[Missing one]] appeared as identical entries. The agent then advised fixing "broken" links first when typing the target would have resolved them (hypothesis: check by adding a type to one such target and re-running validate).
So besides links to somewhere outside the space, validate could say which unresolved targets exist but were excluded, and the setup-space skill could then point at adding type first.
Activity
mindsocket commentedon Sep 27, 2026
Relates to #44 and #80. Solution should consider consistency across these kinds of settings.
Hypothesis: this should be a per space setting, markdown plug-in level if possible, applies to all link fields except hierarchy.
mindsocket commentedon Oct 27, 2026
Another case worth deciding here, from a recorded agent session: a link whose target exists in the space but is excluded during parsing because the note has no
typeis reported under "Unresolved content links (may be outside-space or broken)" alongside genuinely missing targets. In a small test space,[[Untyped target]](an existing, untyped note) and[[Missing one]]appeared as identical entries. The agent then advised fixing "broken" links first when typing the target would have resolved them (hypothesis: check by adding atypeto one such target and re-runningvalidate).So besides links to somewhere outside the space,
validatecould say which unresolved targets exist but were excluded, and the setup-space skill could then point at addingtypefirst.Report unresolved content links at info severity
Resolve content links to files a space reads but excludes
Report unresolved content links at info severity