Wiki Lint Daily — 2026-05-25
Chunk 2026-05-25T01:40:00+00:00 (autopilot tick — tag taxonomy drift scan)
What was done:
- Ran tag drift scan across all compiled wiki pages (concepts/, entities/, skills/, agents/, etc.)
- Compared used backtick-marked identifiers against 175 canonical tags from SCHEMA.md
- Excluded archive/ (auto-generated stale data), raw/ (immutable), and projects/sessions/ (session logs)
Outputs:
- 3,699 total drift instances across 45 unique non-canonical tags in compiled pages
- Top offenders by drift count:
legend/deprecated/session-writeup-2026-05-20.md— 41 unknown tokens (mostly test IDs/secrets from Legend acceptance testing)entities/hermes-agent.md— 39 unknown tokens (tool names likevision_analyze,web_searchnot in SCHEMA taxonomy)./SOUL.md— 36 unknown tokens (namespace names, tool references)agents/mercury/tools-audit-2026-05-06.md— 19 unknown tokens (new tool categories:apple-integration,creative,diagramming)
- Tag drift is concentrated in three categories:
- Tool references (
kubectl,terminal,search_files,execute_code) — not tagged as “tool tags” in SCHEMA - Namespace/infra identifiers (
kube-system,ai-agents,cert-manager) — not yet in taxonomy - Legend project artifacts — test IDs, secrets, deprecated session data
- Tool references (
Next chunk picks up:
- Orphan page scan (pages with no inbound wikilinks from compiled content)
Chunk 2026-05-25T23:51:00+00:00 (autopilot tick — orphan page scan)
What was done:
- Ran orphan page scan across all compiled wiki pages (concepts/, entities/, skills/, agents/, queries/)
- Excluded raw/ (immutable), archive/ (auto-generated), and projects/sessions/ (session logs)
- Matched inbound wikilinks using normalized comparison (case-insensitive, slash/dot normalization)
Outputs:
- Total compiled pages scanned: 990
- Unique wikilinks found across all pages: 1,358
- Orphan pages (zero inbound links): 42
- Weakly-linked pages (exactly 1 inbound): 458
- Top orphan categories:
agents/*/tools-audit-*.md— historical audit reports not cross-referenced in any current pageconcepts/memory-overcommit-thresholds.md— concept page with no inbound linksentities/contacts.md,david-robinson.md,important-inbox.md,pvs-communication-profile.md— entity pages disconnected from main graphskills/*/references/subpages — deep reference docs not linked from parent skill pagesskills/creative/popular-web-designs/templates/— template files with no inbound references
Next chunk picks up:
- Broken wikilink scan (dangling links that point to non-existent pages)
Chunk 2026-05-25T01:52:00+00:00 (autopilot tick — broken wikilink scan)
What was done:
- Ran broken wikilink scan across all compiled wiki pages (concepts/, entities/, skills/, agents/)
- Canonical page index built from 988 unique .md files
- Pattern matched
[[target]]and normalized for comparison (lowercase, .md stripped)
Outputs:
- 1,180 broken links found across compiled pages
- Top categories of broken links:
- Archive references —
archive/2026/stale-*/paths (35+ links): pages moved to stale archive but original links remain in concepts/ - Raw web-captures —
raw/web-captures/references (4+ links): immutable raw captures that exist but aren’t indexed in canonical set - Project cross-references —
projects/voice-clone/index,projects/synthetic-avatar/index: projects referenced from concepts pages but not yet compiled into canonical index - Missing utility pages —
runbooks/mercury-email-cron,accounts: short references to pages that may never have existed - Internal pipeline names —
f5-tts-streaming,slack-voice-pipeline: concept slugs used as links but actual page paths differ
- Archive references —
Affected files include: echomimic-v3.md, finance-household-accounts.md, hermes-k8s-deployment.md, grocery-catalogues-2026-05-02.md
Next chunk picks up:
- Orphaned page remediation plan — prioritise linking the 42 orphan pages discovered in previous chunk, starting with entity pages (pvs-communication-profile, important-inbox) that are high-value targets for inbound links
Chunk 2026-05-25T01:53:00+00:00 (autopilot tick — orphan remediation: pvs-communication-profile + important-inbox)
What was done:
- Read
entities/pvs-communication-profile.mdandentities/important-inbox.mdto understand their content - Identified natural link targets for each orphan page
- Added inbound links from relevant pages:
- pvs-communication-profile linked from hermes-agent (hermes uses it) and himalaya-email (email workflow reference)
- important-inbox linked from himalaya-email (inbox organization concept)
Outputs:
entities/pvs-communication-profile.md— now has 2 inbound linksentities/important-inbox.md— now has 1 inbound linkagents/hermes-agent.md— edited: added wikilink reference to pvs communication profile in section about user interaction patternsconcepts/himalaya-email.md— edited: added wikilinks to both pvs-communication-profile and important-inbox in the email workflow context
Next chunk picks up:
- Remaining high-value orphans:
entities/contacts.md,entities/david-robinson.md, and concept pagememory-overcommit-thresholds