> ## Continuous-process anchor
>
> This spec describes an instance of one of the retired-script themes
> documented in docs/design/retired_scripts_patterns.md. Before
> implementing, read:
>
> 1. The "Design principles for continuous processes" section of that
> atlas — every principle is load-bearing. In particular:
> - LLMs for semantic judgment; rules for syntactic validation.
> - Gap-predicate driven, not calendar-driven.
> - Idempotent + version-stamped + observable.
> - No hardcoded entity lists, keyword lists, or canonical-name tables.
> - Three surfaces: FastAPI + orchestra + MCP.
> - Progressive improvement via outcome-feedback loop.
> 2. The theme entry in the atlas matching this task's capability:
> S1 (pick the closest from Atlas A1–A7, Agora AG1–AG5,
> Exchange EX1–EX4, Forge F1–F2, Senate S1–S8, Cross-cutting X1–X2).
> 3. If the theme is not yet rebuilt as a continuous process, follow
> docs/planning/specs/rebuild_theme_template_spec.md to scaffold it
> BEFORE doing the per-instance work.
>
> **Specific scripts named below in this spec are retired and must not
> be rebuilt as one-offs.** Implement (or extend) the corresponding
> continuous process instead.
Task ID: 5faca020-48e8-45e6-aec0-5c54cda327a0 Type: recurring (daily) Layer: Quality
Sample 10 wiki pages from /wiki. Check: content quality, mermaid diagrams rendering,
cross-links to analyses/hypotheses, infobox data present. Fix formatting issues.
# Title for proper structureSampled pages reviewed:
## Introduction or infobox)Sampled pages reviewed (10 pages, diverse entity types):
[@ref]) leaking into HTML table rows — common in auto-generated infoboxes<td> not rendering (needs <strong> tags)scripts/wiki_quality_review_20260408.py — reusable fix functions for all identified patternsSampled pages reviewed (10 pages, diverse entity types):
scripts/wiki_quality_fixes_20260410.py contains the correct content for companies-boston-scientific for future application once DB WAL state is repaired.New issue categories identified:
mermaid blocks before H1 in AI tool pages (auto-generated content precedes structured header) — contains corrected content for all 6 fix pages2026-04-12 — Task 5faca020
Sampled pages reviewed (10 pages, diverse entity types):
- genes-kif26a (gene, 3004 words): trivial word-split mermaid (Kinesin/Family/Member nodes)
- genes-marchf2 (gene, 1912 words): trivial word-split mermaid (Gene/Membrane-Associated/RING-CH)
- genes-tab1 (gene, 1871 words): trivial mermaid with HTML artifact nodes ("class", "infobox" as mermaid nodes)
- proteins-gba1-protein (protein, 1280 words): YAML frontmatter (
---title: ...---) leaked into body; duplicate H1 mid-page
proteins-syndecan3 (protein, 918 words): first infobox showed intervention data (not protein data); duplicate H1 mid-page; boilerplate intro
companies-paxos-therapeutics (company, 620 words): missing infobox → added minimal one
mechanisms-mtor-signaling-cbs-psp (mechanism, 1561 words): good quality, no fixes needed
proteins-igf2bp1-protein (protein, 1584 words): good quality, no fixes needed
genes-dysf (gene, 2154 words): good quality, no fixes needed
genes-ctsw (gene, 2039 words): good quality, no fixes needed
Fixes applied:
Bulk fix detail:
Scanned all 16,189 wiki pages with mermaid content. Found 5,016 pages with trivial
word-split mermaid diagrams — autogenerated blocks that split a gene/protein name into
single-word nodes connected by "related to" edges. No biological meaning. All removed.
Detection criteria: all edges are "related to", satellite nodes have numbered suffix pattern
(GENE_N["Word"]), all labels are single words.Scripts:
fix_wiki_quality_5faca020.py (page-level fixes), fix_trivial_mermaids_batch.py (bulk fix)New issue categories identified:
- Trivial word-split mermaids (now fixed at scale for all 5,016 affected pages)
- YAML frontmatter leaking into markdown body content
- Malformed infoboxes showing content-section data instead of structured entity metadata
2026-04-12 — Task 5faca020 (second run)
Sampled pages reviewed (10 pages, diverse entity types):
- proteins-fgfr3 (protein, 574 words): good quality, no issues
- genes-s100a6 (gene, 829 words): mermaid OK, infobox OK
- mechanisms-phase-3-trial-readiness-matrix (mechanism, 1484 words): broken self-referential mermaid
- genes-tlr5 (gene, 1563 words): citation leaks in HTML infobox, boilerplate intro
- mechanisms-neurogenesis-4r-tauopathies (mechanism, 2808 words): citations in prose only (correct usage), no fixes needed
- cell-types-accessory-olivary-nucleus (cell, 872 words): duplicate H2, broken mermaid, placeholder "Taxonomy|ID" infobox row, boilerplate intro
- cell-types-bergmann-glia (cell, 1029 words): markdown bold in HTML label cells, duplicate Cell Ontology row
- mechanisms-contradiction-detection-neurodegeneration (mechanism, 1428 words): good quality, no fixes needed
- proteins-eif2s1-protein (protein, 570 words): raw CSS leaked into markdown, MediaWiki infobox syntax inside HTML div, boilerplate intro, filler Background section, duplicate References sections (×4)
- cell-types-lingual-cortex-neurons (cell, 614 words): broken mermaid, placeholder infobox rows (Taxonomy/Marker headers), boilerplate intro
Fixes applied (6 of 10 pages had issues):
Script: scripts/wiki_quality_fix_2026_04_12.pyNew issue categories identified:
- Raw CSS stylesheet blocks leaked into markdown content (likely from broken template rendering)
- MediaWiki wikitable syntax (
, ! Header, | Cell) inside HTML div wrappers
Citation markers appended to end of HTML table row tags ( </tr> [@ref2016])
2026-04-17 — Task 5faca020
Sampled pages reviewed (10 pages, diverse entity types):
- proteins-fyn-protein (protein, 9857 chars): YAML frontmatter leaked into body
- cell-types-medial-vestibular-nucleus-neurons (cell, 7758 chars): broken self-referencing mermaid with HTML class names as node labels
- mechanisms-synaptic-organization (mechanism, ~10K chars): OK (citations in prose are correct usage)
- mechanisms-cortisol-tau-pathway, mechanisms-braak-staging, cell-types-astrocytes-hepatic-encephalopathy, etc.: flagged as "malformed_mermaid" but were false positives (valid mermaid
style directives)
Issues investigated but found to be false positives:
- "malformed_mermaid" detection was incorrectly flagging pages with valid
style declarations in mermaid blocks (e.g., style H fill:#0e2e10,stroke:#333) - these are valid mermaid syntax
"citation_leak" detection was incorrectly flagging citations in prose (e.g., [@bellenchi2021] at end of sentence) - these are correct usage
Fixes applied (2 of 10 pages had real issues):
Script: scripts/wiki_quality_fix_5faca020.pyDB contention note: Direct SQL updates to PostgreSQL required multiple retries due to WAL mode lock contention from running API service. Updates succeeded on retry with 180s timeout and PASSIVE checkpoint.
2026-04-20 — Task 5faca020
Sampled pages reviewed (10 pages, diverse entity types):
- clinical-trials-urolithin-a-parkinsons-nct06033890 (trial, 16492 chars): missing H1, YAML-like frontmatter at start
- cell-types-calcium-dysregulated-neurons (cell, 19358 chars): citation leaks in table infobox, duplicate H1, filler Background
- cell-types-nucleus-ambiguus-cardiac-control (cell, 11031 chars): citation leaks in div infobox, filler Background
- proteins-dync1i1-protein (protein): filler Background
- therapeutics-mas-receptor-agonists (therapeutic): markdown bold in HTML cells
- genes-tlr10 (gene): filler Background
- proteins-elovl7-protein (protein): citation leaks, filler Background
- genes-map1lc3b (gene): markdown bold in HTML cells
- researchers-giovanni-volpe (researcher): duplicate H1 ("# Overview" then "# Giovanni Volpe")
- proteins-cofilin-1 (protein): citation leaks, filler Background
Fixes applied (10 of 10 pages had issues):
Scripts: 2026-04-20 (evening) — Task 5faca020
Pages audited: 300+ wiki pages across ~20 random samples of 30 pages each.
Issues found:
- Self-referencing mermaid diagrams (node → itself via "related to")
- Missing H1 headings (page starts with
or infobox div)
Greek unicode in mermaid (α, β, γ, δ, κ, μ)
Trivial word-split mermaid (node IDs splitted into words as separate nodes)
Fixes applied (57 pages total):
Scripts: fix_wiki_quality.py, fix_wiki_batch2.py, fix_wiki_batch3.py, wiki_quality_audit.pyNotes:
- All DB writes via
(PostgreSQL)
Used journaled_update() pattern with direct SQL (%s placeholders) + NOW()`
{
"requirements": {
"analysis": 6,
"reasoning": 6,
"safety": 9
},
"completion_shas": [
"6f71b1ea26612dee439a47b8214c9801ee0ac3e5",
"27a8518f848fb030bb7bb4d1bb305effa0274157"
],
"completion_shas_checked_at": "2026-04-12T18:10:28.083646+00:00",
"completion_shas_missing": [
"c8f8e0d69f5425f216c9f2ea1f6e1398965393a9",
"7341d13c4a2186662531537ddec0dc41207b98ae",
"2259b123c3d7fc1ffce4e7e5bede75087c258aad",
"e51596ca0d02c7244d20af69a8cfe9972ff78278",
"309667156ce072bfa6ea114714bc568cd3b9299b",
"d9d389bade4b623203448206b67f81f272cafc78",
"fe4109ecc5c8638860f85978ba05edfd8efa4e12"
]
}