lint

log/lint.md — lint-pass entries

Sub-file of log — see parent for index.

[2026-06-02] lint — stale page-existence gap sweep

User-prompted: many #gap/needs-page/#gap/needs-stub/#gap/needs-*-page/#gap/needs-seeding/#gap/missing-page markers were placed before the referenced page was seeded and are now stale. Swept the full page-existence gap family (excluding the append-only log/ audit trail), cross-checked every referenced wikilink target against the live file inventory, and removed/rewrote the markers whose target now exists (resolving by basename, so links already pointed to the right file regardless of folder).

Stale gaps resolved (target now exists) — 14 files edited:

• biomarkers/il-6-biomarker.md — il-6 (needs-protein-page), tocilizumab (needs-compound-page)
• studies/iwata-2025-ewingella-americana-antitumor.md — ewingella-americana, bacterial-cancer-therapy (both needs-seeding)
• molecules/compounds/setmelanotide.md — peptide-therapeutics (needs-stub)
• molecules/compounds/tazarotene.md — retinoids (needs-page, ×2)
• interventions/gene-therapy/crispr-base-editing-pcsk9.md — pcsk9 (needs-stub)
• hypotheses/translation-failure-of-aging-interventions.md — interventions-testing-program (missing-page, ×2; reworded body + dropped Limitations bullet)
• molecules/proteins/fgf23.md — bone + osteocytes (needs-page, ×2)
• molecules/proteins/osteopontin.md — bone (needs-page)
• molecules/proteins/runx2.md — matrix-gla-protein, osteoporosis, bone (needs-page)
• cell-types/osteocytes.md — osteoporosis, runx2, bmp-2, osteopontin, bone (needs-page)
• phenotypes/osteoporosis.md — sost (needs-page, ×3)
• molecules/proteins/matrix-gla-protein.md — bmp-2 (needs-page)
• molecules/proteins/sost.md — dkk1 (needs-page)
• pathways/wnt-beta-catenin.md — sost, dkk1 (needs-page)
• molecules/proteins/lrp5-lrp6.md — bone, osteocytes (reworded so the gap attaches only to still-missing osteoblasts)
• tissues/bone.md — osteoporosis (needs-page, ×3)
• tissues/arteries.md — endothelial-cells (needs-page, ×2)
• cell-types/vsmc.md — arteries (needs-page, frontmatter comment + body + See-also)
• molecules/proteins/caspase-1.md — ketogenic-diet (split from still-pending beta-hydroxybutyrate)
• processes/vascular-calcification.md — runx2, bmp-2, osteopontin, wnt-beta-catenin (was wnt-signaling), matrix-gla-protein, phosphate-additive-reduction (needs-page, 7 markers)

Remaining real page-existence gaps, ranked by inbound demand (distinct linking pages):

Rank	Missing page	Inbound	Type / home	Notes
1	osteoblasts	9	cell-type	Highest-demand missing node; master bone-formation cell referenced from runx2/sost/dkk1/lrp5-lrp6/osteocytes/osteoporosis/bone. Seed first.
2	osteoclasts	4	cell-type	RANKL/OPG resorption arm; pairs with osteoblasts.
2	hyperphosphatemia	4	phenotype/node	Central driver node in vascular-calcification causal graph.
4	vitamin-k	4	compound	Substrate for MGP carboxylation; ties to MK-7 supplementation hypothesis.
5	splicing-dysregulation / alternative-splicing	3	process	Canonical home for age-related spliceosome biology contextualizing SF3B1 (3 markers on sf3b1.md). Pick one canonical slug.
6	bmsc (marrow stromal)	2	cell-type	Osteoblast/adipocyte precursor; lineage-drift in bone aging. Partial overlap with existing mesenchymal-stem-cells — decide split vs alias.
6	ctnnb1 (β-catenin)	2	protein	Downstream Wnt effector; demanded by wnt-beta-catenin + lrp5-lrp6.
6	cardiovascular-system	2	framework MOC	Organ-system overlay; arteries/heart/myocardium currently un-MOC’d.
6	beta-hydroxybutyrate	2	compound	NLRP3-suppressing ketone body; ketogenic-diet sibling.
6	aim2	2	protein	DNA-sensing inflammasome platform.
11	macrophages	2	cell-type	Surprisingly low inbound — likely under-linked rather than truly low-priority; worth seeding given centrality to inflammaging.
12	romosozumab	1	compound	FDA-approved anti-sclerostin mAb; osteoporosis cluster.
12	nlrc4	1	protein	Alternative inflammasome platform.
12	ggcx, vkor	1 ea	protein	Vitamin-K cycle enzymes; MGP carboxylation pathway.
12	nitric-oxide	1	process	eNOS→NO signaling; endothelial aging.
12	adventitial-fibroblasts, aortic-aneurysm	1 ea	cell-type / phenotype	Arteries-page cross-reference completion.
18	tocotrienols	0	compound	No inbound demand yet; lowest priority (scope-note placeholder on alpha-tocopherol only).

Class-seeding gaps (not single-page; in frameworks/interventions-by-hallmark.md) — left in place, genuinely unseeded: partial-reprogrammer, HDAC-inhibitor, SIRT6-activator, chaperone-enhancer compound pages; TA-65 compound page; AAV-TERT intervention page. Plus a tagging review (spermidine/rapamycin → loss-of-proteostasis cross-link).

log/ audit-trail files were intentionally not edited (historical record).

[2026-05-07] lint

Structural + coverage lint pass run from a cold start; the focus was hallmark-causality coverage and mechanistic depth per hallmark.

Orphans (27 total, triaged):

• 6 framework MOCs never wikilinked from anywhere: frameworks/hallmark-causality-graph.md, frameworks/interventions-by-hallmark.md, frameworks/intervention-classes.md, frameworks/interventions-by-modality.md, frameworks/causal-graph-data.md, frameworks/biological-age-measurement.md. All exist (R14–R16 vintage) and are referenced via path-strings in narrative text but never via [[wikilink]]. Fix: add to index.md § “Frameworks (navigational overlays)” — currently shows only [[hallmarks-of-aging]] and [[sens-damage-categories]].
• 2 biomarker pages orphaned (biomarkers/telomere-length-leukocyte.md, biomarkers/frailty-index.md) — should be linked from frameworks/biological-age-measurement.md (which is itself orphaned) and from the telomere-attrition / phenotypes/frailty pages.
• 7 study pages orphaned (lopez-otin-2013, lopez-otin-2023, soengas-2001, yousefzadeh-2018, tyner-2002, waziry-2023, li-2023) — script alias-blindness is partly responsible (these are actually wikilinked as [[studies/...]] from atomic pages but the basename-only orphan check misses the path-prefixed form). Confirm pattern, then update the lint script to grep both bare-name and path-form patterns.
• 2 atomic-page orphans needing real cross-reference work: molecules/proteins/lamp2.md (only mentioned via the lamp-2a alias from tau.md; main protein page never directly wikilinked), molecules/compounds/empagliflozin.md (clinical-stage:fda-approved, hallmarks:[deregulated-nutrient-sensing, chronic-inflammation, mitochondrial-dysfunction], but no inbound link from those hallmark pages or from the GLP-1/SGLT2 intervention class page).
• 1 hypothesis orphan (hypotheses/negligible-senescence.md) — should be linked from hallmarks/cellular-senescence.md and from any model-organism page covering naked mole rat / hydra.
• 1 pathway orphan (pathways/mismatch-repair.md) — should be linked from hallmarks/genomic-instability.md and from the DNA-damage-response pathway page.
• 1 system orphan (.claude/agents) — expected; ignore.

Causal-graph integrity:

• All 12 hallmarks have populated causes: and caused-by: frontmatter except deregulated-nutrient-sensing.caused-by: which is intentionally empty (per hyperfunction theory framing).
• One bug: hallmarks/telomere-attrition.md causes: lists [[telomere-attrition]] itself (self-loop). Fix on next propagation pass.
• Two cross-edges asymmetric: chronic-inflammation lists stem-cell-exhaustion as both upstream and downstream, and stem-cell-exhaustion does the same back. The mutual-causation framing is biologically defensible (inflammation drives stem-cell exhaustion which drives more inflammation) but the causal-graph framework page should explicitly mark these as feedback loops rather than node-edge violations.

Per-hallmark mechanistic depth (proteins + pathways + processes referencing the hallmark):

Hallmark	Proteins	Pathways	Processes	Compounds/Interventions	Tractability	Status
cellular-senescence	52	17	8	18	high	well-covered
deregulated-nutrient-sensing	50	6	2	21	high	well-covered
disabled-macroautophagy	43	4	6	11	high	well-covered
chronic-inflammation	41	11	5	31	high	well-covered
loss-of-proteostasis	33	3	6	16	moderate	well-covered
mitochondrial-dysfunction	28	9	7	22	moderate	well-covered
genomic-instability	23	13	2	7	low	mech-deep, intervention-thin
stem-cell-exhaustion	21	10	2	20	low	mech-thin, intervention-deep
epigenetic-alterations	12	2	1	7	moderate	severely under-covered
altered-intercellular-communication	5	5	1	15	moderate	mech-thin
dysbiosis	4	1	0	4	moderate	severely under-covered
telomere-attrition	4	0	0	2	low	severely under-covered

Top implicit-stub mechanism clusters (inbound counts via Step-3 discovery, cluster-grouped):

• Mitochondrial dynamics: mfn1 (17), mfn2 (17), mitochondrial-dynamics-pathway (9), miro2 (6) — explicitly listed as planned in mitochondrial-dysfunction verified-scope line.
• Telomere biology: telomerase-pathway (14), telomerase (1), replicative-senescence (1), mtdna (7), dkc1 (7).
• DNA repair detail: pcna (13), xrcc1 (10), mutyh (8), ape1 (8), lig1 (6), polb (5), rpa (6), xpf (5).
• Inflammasome/oxidative: nlrp3 (12; protein, distinct from existing nlrp3-inflammasome pathway page), oxidative-stress (8), reactive-oxygen-species (5), sting (6), irf3 (7).
• CMA/proteostasis detail: lamp-2a (11), chop (9), gadd34 (8), hri (5), wipi2 (5).
• Epigenetic detail: hdac (10), dna-demethylation (6), mitf (7).
• Stem-cell signaling: tgf-beta-pathway (8), wnt-pathway (5), nanog (9), gsk3b (9), smad4 (9), fgf-signaling (7), melanocyte-stem-cells (5).
• Dysbiosis: gut-barrier (10), gut-microbiome (9).
• Heavyweight orphan studies: studies/lu-2020-osk-vision-restoration (10), studies/yang-2023-epigenetic-information-loss (9), studies/ocampo-2016-partial-reprogramming (7), studies/horvath-2013-epigenetic-clock (7), studies/guerreiro-2013-trem2-ad (5).

Intervention coverage gaps surfaced:

• genomic-instability (only 7 inbound interventions despite 23 proteins, 13 pathways) — DNA-repair-augmenting therapeutics largely missing (PARP inhibitors as senolytics, NAD/PARP-axis interventions, DNA-PKcs modulators).
• telomere-attrition (only 2 inbound interventions) — TA-65, telomerase mRNA (Andrews/Idoyaga), GRN163L/imetelstat, AAV-TERT (exists) absent.
• epigenetic-alterations (only 7 inbound interventions) — HDAC-inhibitor class page absent; partial-reprogramming + Yamanaka-factor interventions partly covered (aav-osk exists) but the broader class is thin.
• dysbiosis (only 4 inbound interventions) — FMT, Akkermansia supplementation, prebiotic class, postbiotics absent.

Schema-clarification carry-forwards (none new this pass; all R26+ items already in CLAUDE.md).

Notable surprises:

• The wiki’s mechanistic depth is sharply bimodal: the top 6 hallmarks have 28–52 linked proteins each; the bottom 4 have ≤12 each. The bottom 4 (epigenetic-alterations, altered-intercellular-communication, dysbiosis, telomere-attrition) are also the hallmarks where mechanistic causes are most actively contested in the field — coverage thinness reflects field-uncertainty as much as wiki-incompleteness.
• The stem-cell-exhaustion paradox: 20 inbound interventions despite tractability:low. Most of those interventions target downstream readouts (sarcopenia, immunosenescence) rather than upstream cause. The hallmark page should disambiguate “intervention prevents/reverses upstream cause” from “intervention compensates for downstream consequence.”

[2026-05-04] lint — implicit-stub queue refresh

Ran lint-pass § Step 3 (implicit-stub discovery script) post-Round-4 to re-rank Round 5+ priorities.

Top of queue (≥5 inbound refs after collapsing path duplicates):

• 13 foxo3 (PI3K/IIS) — appears as both foxo3 and molecules/proteins/foxo3 (~7+7-1)
• 10 bim, 10 bad, 7 bid (apoptosis BH3-only)
• 9 beclin-1 (autophagy core)
• 9 spermidine (NAD/autophagy compound)
• 8 akt (PI3K/IIS)
• 7 alzheimers-disease (disease entity)
• 6 each: ubiquitin-proteasome-system, neurodegeneration, navitoclax, mitochondrial-biogenesis, inflammaging, a1331852
• 5 each: pink1-parkin-pathway, pi3k, nampt, deptor, cgas-sting, bone-marrow

Total tier-A entries (≥5 refs): 20 stubs Total tier-B entries (4 refs): 16 stubs Total tier-C entries (3 refs): 22 stubs Total tier-D entries (2 refs): ~50 stubs Total tier-E entries (1 ref): ~200+ stubs (mostly study pages)

Re-prioritization: Original Round 4 plan (phenotypes/disease) was deferred (became actual Round 8) because Round 4 became “model organisms + cell types.” Original Round 5 (hypotheses) deferred to new Round 9. New Round 5 = apoptosis machinery completion (BH3-only family + executioners + senolytic compounds; ~45+ collective refs — single highest-density unblock per page). See ROADMAP.md § “Suggested seeding order.”

Housekeeping issues surfaced (NOT addressed in this lint pass):

• Duplicate study slugs: ~9 instances of the same paper linked under 2-3 different filenames (e.g., willcox-2008-foxo3 vs -foxo3a vs -foxo3a-longevity; holzenberger-2002- vs -2003-; zhu-2015-senolytic-drugs vs -achilles-heel-senescent-cells). Will require a find-and-replace housekeeping pass to normalize.
• Capitalization bugs: Rapamycin, Disabled-macroautophagy
• Trailing-character bugs: senolytics.md (.md suffix in link), stem-cell-therapy/ (trailing slash), bare interventions/

Skipped lint steps in this pass:

• Step 1 (orphan pages, broken links beyond implicit stubs, missing frontmatter) — not run
• Step 2 (unsourced claims sweep, contradictions sweep, stale claims, decayed URLs) — not run; Round 4 verifiers already surfaced and addressed many of these inline
• Step 4 (verify AI-extracted pages) — Round 4 was effectively this for the 12 new pages; remaining verified: false candidates are mostly closed-access study pages awaiting OA access
• Step 5 (gaps tracker) — not refreshed

Recommend a fuller multi-step lint pass after Round 7 closes (when the apoptosis + autophagy + IIS clusters are complete and the link graph stabilizes).

[2026-05-04] lint — full pass post-Round-9

Per sops/lint-pass.md, ran structural + content + coverage + verification lint after Rounds 5-9 closure.

Coverage summary

• Total wiki.md files: 159 (excluding.obsidian,.claude, README)
• verified: true: 129 — 81% of all pages are PDF-verified
• verified: false: 1 — only studies/li-2023-dual-mtor-nad-gerotherapy.md remains
• No verified field (hallmark stubs, MOC pages, frontmatter-only pages): 19

Structural lint

• Orphans (true): 1 — interventions/pharmacological/senomorphics.md is genuinely orphaned (not linked from any parent MOC). Recommend: link from interventions or chronic-inflammation hallmark page when expanded.
• False-positive orphans (regex didn’t match path-form references): .claude/agents/* (agent definitions, not wiki content); studies/* (referenced via [[studies/name]] paths); hypotheses/negligible-senescence.md (referenced via [[hypotheses/negligible-senescence]]); sops/writing-hypothesis-pages.md (referenced from CLAUDE.md).
• Pages missing type: frontmatter: 2 — CLAUDE.md (intentional; not a wiki entity) and log.md (intentional; chronological audit trail). Both correct.

Content lint — gap-tag census

Gap tag	Count
#gap/no-fulltext-access	669
#gap/needs-replication	515
#gap/unsourced	471
#gap/needs-human-replication	406
#gap/no-mechanism	298
#gap/contradictory-evidence	165
#gap/needs-canonical-id	137
#gap/long-term-unknown	128
#gap/dose-response-unclear	93

The no-fulltext-access count reflects that ~25% of cited primary sources are paywalled/closed-access. The unsourced and needs-replication counts will decrease as study pages get filled in (currently only 6 study pages exist for ~50+ referenced studies).

Coverage lint — refreshed implicit-stub queue

647 total broken wikilinks (entities referenced but no page exists). Top-priority gaps after filtering studies/aliases:

Tier A (10+ inbound, high-leverage gaps):

• [[ubiquitin-proteasome-system]] (17) — major hallmark-process; foundational
• [[tbk1]] (15) — TBK1 kinase; phosphorylates OPTN-Ser177 + NDP52 for mitophagy/xenophagy; ALS gene
• [[14-3-3]] (14) — universal phospho-binding scaffold; sequesters BAD, FOXOs, TFEB
• [[xiap]] (12) — IAP-family caspase inhibitor; modulates apoptosis decision
• [[bcl-w]] (12) — BCL-2 family member; IMR90 SCAP component (Yosef 2016 verified)
• [[atg10]] (12) — autophagy E2 for ATG12-ATG5 conjugation
• [[grb2]] (10) — RTK adapter; RAS-MAPK pathway from INSR/IGF1R
• [[brain]] (10) — anatomical reference; tissue stub for cross-linking
• [[bone-marrow]] (10) — anatomical/HSC niche

Tier B (5-9 inbound):

• [[smac-diablo]] (9), [[fip200]] (9), [[atg16l1]] (9)
• [[vdac1]] (8), [[mfn1]] (8), [[mfn2]] (8), [[miro1]] (8) — mitochondrial dynamics; PINK1/Parkin substrates
• [[atg3]] (8) — autophagy E2 for LC3 lipidation
• [[skeletal-muscle]] (8) — tissue
• [[caspase-7]] (7) — executioner caspase paralog of caspase-3
• [[nlrp3-inflammasome]] (7) — major hallmark pathway; CANTOS biology
• [[myocardium]] (7), [[glut4]] (7), [[age-1]] (7)
• [[oxphos]] (6), [[mitochondrial-biogenesis]] (6), [[cardiac-fibroblasts]] (6), [[alpha-synuclein]] (6)
• [[nampt]] (5), [[jak-stat-pathway]] (5), [[deptor]] (5), [[cgas-sting]] (5)

Aliases that look broken but resolve correctly:

• [[inflammaging]] (17) → aliased on [[chronic-inflammation]] hallmark page
• [[irs1]] (10) → aliased on [[irs-1]]
• [[bnip3l]] (4) → aliased on [[nix]]

Verification lint

Only 1 page remains verified: false: studies/li-2023-dual-mtor-nad-gerotherapy.md. All other drafted pages have been verified.

Stub pages still active

22 pages tagged #stub. The most consequential are the 12 hallmark MOC pages, which still carry only one-line descriptions despite being heavily cross-linked. Filling these out is a major Round 10 priority — they’re foundational synthesis pages over 100+ verified atomic entities.

[2026-05-08] lint | UniProt drift sweep (R34 backfill)

Ran the full UniProt canonical-DB drift sweep scaffolded in R34 (sops/lint-pass.md § “UniProt canonical-DB drift sweep”). Script saved to /tmp/uniprot-drift.py; ~3 min walltime against ~177 protein pages.

Pages checked: 179 .md files in molecules/proteins/. Script-skipped pages with uniprot: null (foxo-transcription-factors.md = family page, terc.md = ncRNA) — flagged as fetch errors but not actual drift cases (the script tried to GET /uniprotkb/null and the API correctly returned 400). Effective coverage: 177 pages with a populated uniprot: accession.

Drifts found: 13 pages with frontmatter drift, all in HGNC/NCBI-Gene/Ensembl fields. Zero accession-level drift on uniprot: itself.

Breakdown by field:

• hgnc: 11 drifts — 8 numeric-vs-numeric drift (atg10 21073→20315; atg101 25758→25679; bcl-w 993→995; bcl2l13 15769→17164; deptor 24784→22953; fip200 15276→15574; foxo4 7138→7139; ku70-ku80 12830→4055; myostatin 7546→4223), 3 null-fill cases (14-3-3 null→12855; eif2alpha null→3265; gcn2 null→19687)
• ncbi-gene: 1 drift (14-3-3 null→7534, null-fill)
• ensembl: 1 drift (mitofusins ENSG00000171109→ENSG00000116688 — wiki had MFN1’s ENSG but top-level uniprot: is O95140=MFN2; updated to MFN2 ENSG and corrected the misleading inline comment)
• uniprot accession: 0 drifts (consistent with R34 sample finding; UniProt accessions are extremely stable)

All 13 drift cases fixed in-line. No human-judgment cases — all were mechanical ID corrections (HGNC numeric drifts, null-fills from API, ENSG-paralog mismatch on a family page where the top-level uniprot field was unambiguous). Re-run of the script after fixes returned 0 drifts.

Notable patterns:

• HGNC numeric-vs-numeric drift was the single most common type (8/13). Confirms R34 sample lesson: HGNC IDs are the most drift-prone canonical-DB field.
• Two ku70-ku80 / 14-3-3 cases were “subunit-mismatch” drifts on multi-protein family pages — the top-level uniprot: accession identifies a primary subunit but the HGNC/ENSG fields had been populated from a different subunit. Recommend updating the SOP to flag family-page subunit-consistency as a derived check.
• The mitofusins ensembl drift exposed a contradiction between uniprot: O95140 (MFN2) and the inline comment claiming “MFN1 (primary subunit per R14 family-page convention)” — the comment was wrong and the page’s top-level accessions are all MFN2-keyed. Comment corrected during the fix.

No re-verification needed — no uniprot: accession-level drift was found, so no aging-relevant claims tied to a now-deprecated accession need re-checking.

Pages with verified-scope notes mentioning unchecked canonical-DB fields: atg101, bcl-w, bcl2l13, deptor, fip200, foxo4, ku70-ku80, mitofusins, myostatin all carried verified-scope: text saying “canonical-DB identity fields not independently re-checked” — this drift sweep IS that re-check. Verified-scope strings remain accurate after fixes (the unchecked-at-the-time scope is now checked-at-this-date). No edits to verified-scope descriptions needed; the next routine verifier pass can update those strings if desired.