Skip to content

ONB — TFW-41 / Phase D: Glossary and Adapter Sync

Date: 2026-04-20 Author: Executor (AI) Status: 🟠 ONB — Awaiting answers Parent HL: HL-TFW-41 TS: TS__PhaseD__glossary_and_adapters

1. Understanding

Phase D is the closing phase of TFW-41. The task has two independent deliverables: (1) add 14 new terms to glossary.md covering the concepts introduced in Phases A (TS structure gates), B (workflow gates), and C (dimensional analysis), and (2) sync 4 Antigravity adapter files in .agent/workflows/ to match their updated source workflows in .tfw/workflows/. Both deliverables are pure text modifications — no code, no compilation. AC-2 depends on AC-1 (adapters reference glossary terms, so glossary must be written first per the [depends] annotation).

2. Entry Points

File Role Current state
.tfw/glossary.md Target — add 14 terms 197 lines, ends at Project-Specific Terms section
.agent/workflows/tfw-handoff.md Adapter to sync 148 lines — pre-Phase B (missing Step 0, Execution Loops, Pre-RF Gate, ONB answer protocol)
.agent/workflows/tfw-plan.md Adapter to sync 146 lines — pre-Phase B (missing Step 0, Pre-TS Gate)
.agent/workflows/tfw-review.md Adapter to sync 145 lines — pre-Phase B (old step numbering: Step 0 = Select Mode vs Step 1 in source; missing HL §7 Principles check in Judge)
.agent/workflows/tfw-research.md Adapter to sync 129 lines — pre-Phase C (missing dimensional analysis thread in Step 5)
.tfw/workflows/handoff.md Source truth 161 lines — Phase B version
.tfw/workflows/plan.md Source truth 153 lines — Phase B version
.tfw/workflows/review.md Source truth 153 lines — Phase B version
.tfw/workflows/research/base.md Source truth 131 lines — Phase C version

3. Questions (blocking — cannot proceed without answers)

# Question Answer
1 TS §6 says adapters are "exact copies" of source workflows. The only observed difference between source and adapter is the frontmatter (--- YAML block with description: field). Both source and adapters have identical frontmatter. Should the sync preserve the existing frontmatter unchanged (i.e., copy content verbatim including frontmatter), or strip/replace it? {coordinator fills in}

4. Recommendations (suggestions, not blocking)

  1. Glossary section placement: TS §6 suggests two new sections: ## Execution Gates and ## Research — Dimensional Analysis. This is clean and scannable. I will follow this recommendation unless directed otherwise.
  2. Adapter sync method: The cleanest approach for AC-2 is to overwrite each adapter file with the exact source content (preserving frontmatter). This guarantees zero diff rather than a manual merge that could miss something.
  3. Glossary definition for "Alternative": The TS AC-1 list includes Dimension, Configuration Space, Consistency Check, and Surviving Configuration — but does NOT list Alternative as one of the 14 required terms. The HL §4 Phase D Deliverables list does include "Alternative". I will add it as an unprompted bonus only if there are no concerns — alternatively, I flag it here as a recommendation. If the reviewer counts 14 strictly, omit Alternative; if 15 is acceptable, include it for completeness.

5. Risks Found (edge cases, potential issues not in TS)

  1. "Alternative" term gap: HL §4 Phase D Deliverables lists "Alternative" (14th item in the deliverables list), but TS AC-1 lists only 14 terms and does NOT include "Alternative" in the explicit checklist. The TS checklist is the contract. I will follow the TS checklist (14 terms, no "Alternative") and note this discrepancy here. The reviewer may flag it as a gap.
  2. Adapter line count drift: TS §6 references Phase B RF line counts (handoff: 161, plan: 153, review: 153) and Phase C RF (research/base: 132 lines). The source files I read today are 161, 153, 153, 131 lines respectively — close but research/base.md is 131, not 132. This is within normal (a trailing newline difference). The sync will copy the actual current state, which is authoritative.

6. Inconsistencies with Code (spec vs reality)

  1. TS §6 "exact copy" claim vs adapter state: The current adapter files are definitively NOT exact copies of their sources — they are pre-Phase B/C versions. This is expected (Phase D exists to fix this), but confirms the magnitude: all 4 adapters need full replacement.
  2. tfw-review.md step numbering: The adapter has "Step 0: Select Review Mode" while the source has "Step 1: Select Review Mode" (with a new "Step 0: Name This Session" prepended in Phase B). The sync will correct this automatically via full replacement.
  3. AC-1 count vs HL deliverables: HL §4 lists "Alternative" in the glossary deliverables list, but TS AC-1 does not include it in the 14-term checklist. TS is the contract. Treated as out-of-scope.

7. Knowledge Citations

# HL §7.2 ref Read? Applied / N/A Notes
1 conventions.md §14 — Anti-patterns list ✅ Applied — confirmed no executor role-lock violations in this phase (glossary + adapter copy = permitted executor work)
2 conventions.md §3 — TS definition ✅ Applied — reinforces that TS is the contract; TS AC-1 (14 terms) overrides HL deliverables list discrepancy on "Alternative"
3 glossary.md — Scope Budget ✅ N/A — Phase D is 0 new files, 5 modifications, well within budget
4 README.md Values — "The thinking is the product" ✅ Applied — glossary definitions describe the thinking concepts (WHAT), not implementation (HOW), per P2 principle

Additional PV items found:

# Source Item Relevance
5 conventions.md §14.1 Terminology origin note — GMA/Zwicky terms MUST NOT appear in researcher-facing files Critical constraint for glossary definitions of dimensional analysis terms: definitions must use TFW native terms only

ONB — TFW-41 / Phase D: Glossary and Adapter Sync | 2026-04-20