Skip to content

RF โ€” TFW-42 / Phase B: Workflow Updates

Date: 2026-04-30 Author: Executor (Antigravity) Status: ๐ŸŸข RF โ€” Complete Parent HL: HL-TFW-42 TS: TS Phase B


1. What Was Done

New Files

File Description
(none) โ€”

Modified Files

File Changes
.tfw/workflows/research/base.md Step 0: iterations.yaml โ†’ research/iterations.yaml, researchN/ โ†’ research/iterN/, RES__* โ†’ research/iterN/RES.md. Step 3: research/ โ†’ research/iter1/, researchN/ โ†’ research/iterN/, template list now explicit with numbered names. Step 4: briefing path โ†’ research/iterN/1_briefing.md, template ref โ†’ 1_briefing.md. Step 5: template ref now lists 2_gather.md, 3_extract.md, 4_challenge.md. Step 6: stage file list numbered, RES output โ†’ research/iterN/RES.md, removed iteration-specific naming variants.
.tfw/workflows/plan.md Step 6b: iterations.yaml path โ†’ research/iterations.yaml, added agent + sources optional fields, added 1-sentence multi-agent reference to conventions.md ยง4. Step 6c: RES__* โ†’ research/iterN/RES.md, iterations.yaml โ†’ research/iterations.yaml (3 occurrences). Step 7: PhaseA/ โ†’ phase-a/, PhaseB/ โ†’ phase-b/, artifact filenames updated, removed RES, from tree comment (RES no longer at task root).

2. Key Decisions

  1. plan.md Step 6c also updated (beyond explicit TS scope). Step 6c had 3 references to old-convention paths (RES__*, iterations.yaml at task root). TS ยง7 Definition of Failure explicitly forbids old paths in "either workflow", so updating Step 6c was mandatory to pass the DoF gate. This was flagged in ONB ยง5 Risk #1.

  2. Removed RES, from plan.md Step 7 tree comment. The old comment said โ† master HL, RES, research/ here. Since RES files now live inside research/iterN/RES.md (not at task root), the RES, reference was stale. Updated to โ† master HL, research/ here to match conventions.md ยง4 Multi-phase folder structure.

  3. Explicit numbered file lists in base.md. Where the old workflow said "Copy templates from templates/research/", the new version lists all 4 numbered files explicitly: 1_briefing.md, 2_gather.md, 3_extract.md, 4_challenge.md. This reinforces P2 (sort order = execution order) by making the numbering visible at every reference point.

3. Acceptance Criteria

  • [x] AC-1: research/base.md paths updated
  • [x] Step 0: iteration detection checks research/iterations.yaml (line 17)
  • [x] Step 0: iteration count uses research/iterN/ folders (line 19)
  • [x] Step 0: resume checks use research/iterN/RES.md (lines 24-25)
  • [x] Step 3: iteration 1 creates research/iter1/, iteration N creates research/iterN/ (lines 41-42)
  • [x] Step 3: no reference to research2/, research3/ or researchN/ (lines 41-42)
  • [x] Step 4: briefing path = research/iterN/1_briefing.md (line 53)
  • [x] Step 5: stage references use numbered names (line 64)
  • [x] Step 6: RES output path = research/iterN/RES.md (line 88)
  • [x] Step 6: no reference to RES__{ID}__title.md or RES__iterN__title.md (lines 87-88)
  • [x] Template copy instruction references numbered files (line 44)
  • Gate: grep researchN/, research2/, RES__iter, briefing.md (without number) โ€” 0 matches each โœ…

  • [x] AC-2: plan.md phase structure

  • [x] Step 7 code block shows phase-a/, phase-b/ (lines 137, 140)
  • [x] Artifact filenames: HL__phase-a__{title}.md, TS__phase-a__{title}.md (lines 138-139)
  • [x] research/ location shows research/ at task root (line 136)
  • Gate: 0 matches for PhaseA in plan.md โœ…

  • [x] AC-3: plan.md multi-agent reference

  • [x] Step 6b mentions agent and sources as optional fields (line 95)
  • [x] Step 6b includes multi-agent reference: "For multi-agent research, see conventions.md ยง4 (Agent selection guidance)." (line 97)
  • [x] Step 6b states iterations.yaml location = research/iterations.yaml (line 90)
  • [x] No tool brand names in plan.md โ€” 0 matches for Antigravity/Claude Code/Codex โœ…

4. Verification

  • Lint: N/A (markdown documentation changes only)
  • Tests: N/A (no test infrastructure for conventions)
  • Verify:
  • Grep researchN/ in base.md โ†’ 0 matches โœ…
  • Grep research2/ in base.md โ†’ 0 matches โœ…
  • Grep RES__iter in base.md โ†’ 0 matches โœ…
  • Grep briefing.md (without number) in base.md โ†’ 0 matches โœ…
  • Grep PhaseA in plan.md โ†’ 0 matches โœ…
  • Grep researchN/ in plan.md โ†’ 0 matches โœ…
  • Grep Antigravity|Claude Code|Codex in plan.md โ†’ 0 matches โœ…
  • Visual verification of all updated sections โ€” confirmed correct โœ…

5. Observations (out-of-scope, not modified)

# File Line(s) Type Description
1 .tfw/workflows/handoff.md 140-144 naming Multi-Phase Task Flow example still uses HL__PhaseA, TS__PhaseA, RF__PhaseA, HL__PhaseB, TS__PhaseB, RF__PhaseB. Needs kebab-case update. Carried from TD-112
2 .tfw/compilable_contract.md 56, 78 naming Still references PhaseA/ in resolution rules. Carried from TD-111

6. Fact Candidates

No fact candidates.

7. Strategic Insights (Execution)

No strategic insights.

8. Diagrams

No diagrams.


RF โ€” TFW-42 / Phase B: Workflow Updates | 2026-04-30