Briefing¶

Parent: HL-TFW-41 Goal: Add structural gates to every critical TFW handoff point so that failures are prevented, not just detected.

Research Plan¶

Mine Helpdesk HD-9 through HD-18 task artifacts (TS, RF, REVIEW, ONB) for concrete evidence of problems #1-9 from HL §2
Extract TS size/content metrics across HD-16 phases (A-D) — verify the inverse correlation claim (S2)
Collect examples of coordinator plan≠fact drift from HD-18 artifacts
Search external sources: "requirements-driven task specifications", "quality gates in agile", "execution loops in engineering workflows"
Research Zwicky's Morphological Box: original methodology, modern applications, AI/research context applicability

Build parameter matrix: gate type × enforcement point × verification method × token cost
Apply Zwicky Box to TFW-41's own solution space as a live test of H4
Compare Requirements-first TS pattern against industry alternatives (BDD/Gherkin, ATDD, Shape Up pitches)
Evaluate token/word budget impact of proposed gates on existing workflows (handoff.md, plan.md, review.md)

Stress-test Execution Loops against real HD-16 Phase C: would they have caught per_page=100 vs spec 500?
Stress-test Pre-TS Gate against HD-18: would reading RF Phase N-1 have prevented the 3 errors?
Counter-evidence: when do gates ADD overhead without catching errors? (false positive rate)
Challenge Zwicky Box: is it genuinely useful for AI research or is it overhead that a good researcher does implicitly?

#	Hypothesis	HL Status
H1	Requirements-first TS reduces "Phase D cleanup" phases	open
H2	Pre-TS Gate (read RF N-1) eliminates plan≠fact drift errors	open
H3	Execution Loops catch more issues than linear execution	open
H4	Zwicky's Morphological Box in research Extract/Synthesis improves decision quality by making researchers systematically consider parameter combinations instead of picking first viable option	open

In scope: Evidence from Helpdesk project (HD-9..HD-18). External research on quality gates, requirements-driven specs, Zwicky Box. Token budget analysis for proposed workflow changes. TS template structure analysis.
Out of scope: Implementation of changes (that's execution). Other projects beyond Helpdesk for evidence. Changes to research workflow beyond Extract/Synthesis (per HL §10 note on H4).

The Helpdesk project is at d:\projects\research\helpdesk\ — should I mine ALL task folders (HD-9 through HD-18) or focus on the specific cases cited in the HL (HD-9, HD-16, HD-18)?
For H4 (Zwicky Box): is the intent to add it as a mandatory tool in research Extract, or as an optional technique the researcher MAY use?
Are there other live projects beyond Helpdesk where you've observed the same 9 problems?

Focus on cited cases only: HD-9, HD-16, HD-18. Don't mine all HD-9..HD-18.
Zwicky Box = mandatory tool in research Extract stage (not optional).
TFW project itself may have evidence of the same problems — check TFW tasks too.

Stage complete: YES