trace-first-starter

ONB — TFW-24: Researcher Role & RES State Machine

Date: 2026-04-04 Author: Executor Status: 🟠 ONB — Awaiting answers Parent HL: HL-TFW-24 TS: TS__TFW-24


1. Understanding

Extract the Researcher as a 4th standalone role (alongside Coordinator, Executor, Reviewer) with its own Role Lock. Introduce a research/ subfolder convention where each research stage writes its own file — file existence = state completion. Add Step 0 (Resume Protocol) to base.md. Convert RES.md template from stage-based to synthesis-based format. Update HL.md template §1 to Vision/Impact/Quote and §2/§5 to domain-agnostic phrasing. Update plan.md with Researcher handoff + Hard Stop. Sync 3 adapter files.

2. Entry Points

File Key Lines/Sections
.tfw/conventions.md §4 (naming), §8 (workflows table), §15 (role lock table)
.tfw/glossary.md Coordinator role definition (~L97-101), Roles section (~L89-118)
.tfw/workflows/research/base.md L7-8 (Role Lock), L12+ (Steps 1-7)
.tfw/workflows/plan.md L64-74 (Step 6: RESEARCH decision)
.tfw/templates/RES.md L50-102 (Stage sections), L127-155 (Closure)
.tfw/templates/HL.md L9-12 (§1), L14-16 (§2), L42-44 (§5), L79+ (§10)
.tfw/PROJECT_CONFIG.yaml L73-76 (RES status)

3. Questions (blocking — cannot proceed without answers)

# Question Answer
No blocking questions

4. Recommendations (suggestions, not blocking)

  1. TS Step 4 (RES.md) says “Remove stage sections… Replace with synthesis structure” but lists 6 sections. Current RES.md has Briefing, Hypotheses, Decisions, Open Questions, 3 Stage sections, Final Checkpoint, Closure (with HL Recommendations, FC, Conclusion). TS’s synthesis structure = Decisions, Open Questions, Hypotheses, HL Update Recommendations, Fact Candidates, Conclusion. I will keep the header + Research Context + Briefing reference as TS instructs (L148: “Keep header… Briefing section stays in RES”).

  2. TS HL §1 uses “Vision” as heading but existing TFW-24 HL uses “Goal & Value”. The template change is to the template not retroactive — understood, will apply to templates/HL.md only.

  3. TS Step 2 specifies Step 4 Briefing → research/briefing.md and Step 5 stages → individual files, but the OODA Stage Loop and Stage Checkpoint sections in base.md reference “Update RES” (L50) and “Update RES — Decisions and Open Questions” (L55). After the change, these should reference stage files instead. I will update the OODA ACT step and Stage Checkpoint accordingly.

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

  1. Word count tight. base.md is currently 518 words (by my count of the 103-line file). Adding Step 0 (~30 words), subfolder instruction (~30 words), and Synthesis rewrite (~40 words) = ~618. May need to tighten existing text slightly to stay ≤600.

  2. Three-adapter sync. TS lists 3 cp commands. .agent/workflows/tfw-research.md is a copy of research/base.md, .agent/workflows/tfw-plan.md and .claude/commands/tfw-plan.md are copies of plan.md. All three currently identical to their sources — confirmed by reading.

6. Inconsistencies with Code (spec vs reality)

  1. TS §4 Step 1 says conventions.md §4 (Artifact Naming) — in the actual file, §4 is titled “Task Numbering” (L74). Research subfolder convention fits here as an extension of the naming/folder rules. No conflict, just noting the title difference.

*ONB — TFW-24: Researcher Role & RES State Machine 2026-04-04*