Skip to content

title: "Gather — "What do we NOT know?"" source: "tasks/TFW-27__wiki_polish_and_brand/research/gather.md"


Gather — "What do we NOT know?"

Parent: HL-TFW-27 Goal: Transform raw documentation output into a polished, branded, navigable knowledge product.

Findings

G1: OSS Methodology Landing Page Patterns

Analyzed 4 projects: Conventional Commits, Keep a Changelog, SemVer, OpenClaw.

Key pattern: TFW is a hybrid (methodology + tool). Methodology projects lead with spec/FAQ. Product projects lead with quick start. TFW needs both.

Proposed landing structure: logo → tagline ("Traces Over Code") → value proposition → core values (3 cards) → quick start (3 steps) → what's inside (link map) → badges → task board (README only).

G2: Dynamic Navigation (MkDocs)

Problem (TD-71): Knowledge topics hardcoded in mkdocs.yml. New topic = manual edit.

Solution: mkdocs-literate-nav plugin + mkdocs_gen_files.Nav() API. gen_docs.py generates SUMMARY.md programmatically: - Static sections (Home, Getting Started, Concepts, Reference) — hardcoded per §16.4 - Dynamic sections (Knowledge Topics, Tasks, Workflows, Templates) — filesystem scan - Hardcoded nav: removed from mkdocs.yml

~20-30 LOC addition to gen_docs.py. Same plugin ecosystem (gen-files author = literate-nav author).

Problem A — 404 links: Existing markdown links [text](path) that break after compilation because source→output paths differ. ~31 instances in .tfw/ sibling links and README→.tfw/ links. Fix: path rewriter using Source Manifest mapping.

Problem B — Missing links (Wikipedia-style): Plain text mentions of known entities that SHOULD be clickable links but aren't. Examples: - D24 mentioned as text → should link to KNOWLEDGE.md decisions table - [TFW-18](../../TFW-18__knowledge_consolidation/HL-TFW-18__knowledge_consolidation.md) mentioned without artifact type prefix → should link to task folder - P7 mentioned → should link to KNOWLEDGE.md principles - F4 mentioned → should link to knowledge topic file - .tfw/workflows/plan.md in prose → should link to compiled page

gen_docs.py has 6 resolvers (of 9 defined in §16.2). Missing: P{N}, F{N}, S{N}. Also missing: bare task IDs without type prefix, file paths without backticks.

Goal: Maximum link density — like Wikipedia, every mention of a known entity becomes a clickable link on the compiled site. Build-time only (raw markdown stays as text references per §16.2 convention).

G4: Artifact Graph — Decision: NOT needed

User decision after discussion: No separate graph.yaml file. Reasons: 1. TFW's markdown files (KNOWLEDGE.md, reference patterns) already encode the relationship graph 2. Compiled docs with Wikipedia-linking = navigable graph for humans 3. AI agents can navigate TFW's markdown directly — TFW is designed for this 4. Don't create unnecessary entities (TFW principle)

Implementation: gen_docs.py builds an in-memory entity index at build time by scanning KNOWLEDGE.md, TECH_DEBT.md, tasks/. Uses it for Wikipedia-style link resolution. No file on disk.

G5: Graph/Schema Standards

No existing standard fits (JSON-LD, RDF, CITATION.cff all wrong scope). Moot — graph.yaml deferred. In-memory entity index in gen_docs.py needs no schema standard.

Checkpoint

Found Remaining
Landing page: hybrid methodology+tool pattern None
Dynamic nav: literate-nav + gen_docs.py Nav() None
Link quality: 2 problems (404 paths + missing Wikipedia links) Need to map ALL entity types for resolver in Extract
Graph: not needed (in-memory index only) None

Sufficiency: - [x] External source used? (CC, KaC, SemVer, OpenClaw, literate-nav docs) - [x] Briefing gap closed? (all gather topics resolved with user direction)

Stage complete: YES