Source: frameworks/gost-enablement/04-gost-enablement-quality.md

GOST Participant Enablement — Quality

Two-gate quality system. Gate 1 is pre-build (binary pass/fail). Gate 2 is post-build (100-point weighted scoring).

Gate 1 — Pre-Build (All Required Inputs Present)

Run this gate before building the participant document and consultant brief. Any failure stops the build.

[ ] Client's GOST documentation is available (training deck, strategy workshop output, or equivalent)
[ ] Company goal statement identified — verbatim from client source, not paraphrased
[ ] All company objectives identified — complete list with measurable targets
[ ] At least one golden example is available (client-specific completed tactic, or kit default HR example confirmed as appropriate)

[ ] Target quarter confirmed with advisor (e.g., Q2 2026)
[ ] Quarter start and end dates confirmed
[ ] Weekly date ranges calculated for the output template (12 weeks, Monday–Friday)

[ ] Client's 12-Week Tactics Planner template reviewed
[ ] Output format in the prompt matches the client's template structure (section names, table columns, field labels)
[ ] If the client's template differs from the kit default, all differences documented and prompt updated

[ ] Company name — correct spelling confirmed
[ ] Client reference data consulted for any proper nouns in the prompt (department names, tool names, framework terminology)

Gate 1 result: [ ] PASS — proceed to build | [ ] FAIL — resolve gaps before building

Run this gate after building both documents. Fix blocking failures first, then score.

#	Check	Pts
1	Company goal in prompt matches client's verbatim goal statement	5
2	All company objectives in prompt match client's verbatim objectives — none missing, none added	10
3	Objective labels (Obj 1, Obj 2, etc.) are consistent throughout the prompt	2
4	Golden example is present and complete (tactic statement, end state, KPIs, backward plan)	5
5	Golden example content is NOT from the current client's data (it's a teaching example, not a real submission)	3
6	SMART criteria are correctly defined (all 5 elements, no errors)	3
7	Bad/good tactic example pair is present and the distinction is clear	2
8	Quarter dates are correct throughout (prompt body, output template, backward plan month labels)	5

#	Check	Pts
9	Prompt opens with a clear instruction to the LLM (what to do, how to do it)	3
10	Company context block is complete (Goal + all Objectives + Strategy/Tactic definitions)	5
11	All 6 steps are present in order (About You, Pick Tactic, End State, Backward Planning, Reality Check, Produce Planner)	5
12	Step 2 includes push-back instruction ("Push back if too vague, too ambitious, or not connected")	3
13	Step 4 walks backward (final month first, then middle, then first month)	3
14	Step 6 includes the complete output format template with all sections and placeholder labels	3
15	Prompt ends with "Start now... Be friendly and conversational."	3

#	Check	Pts
16	Header fields present: Department, Leader, Strategy, Objective(s) Served	3
17	TACTIC field present with placeholder	2
18	END STATE VISION section: description + 3 KPIs with target placeholders	3
19	BACKWARDS PLANNING CALENDAR: all 3 months present with correct month labels for the target quarter	3
20	Each month has a table with columns: Week, Action, Owner, Hours, Success Criteria, Dependencies	3
21	Weekly date ranges in the tables match the target quarter	2
22	Monthly checkpoints present for all 3 months	1
23	CAPACITY ANALYSIS section present (hours allocated, available, buffer, weekly commitment)	2
24	RESOURCE REQUIREMENTS table present (Skills, Tools, Budget, External)	1
25	DEPENDENCY TRACKER table present (Need, From Who, By When, Risk, Backup Plan)	1
26	WEEKLY PROGRESS TRACKER present with all 12 weeks	1
27	LESSONS LEARNED section present (What Worked, What Didn't, Adjustments)	1

#	Check	Pts
28	Participant directions appear BEFORE the prompt (not after)	3
29	"What this is," "What you'll get," "Time," and "What you need" sections present	2
30	"Have more than one tactic?" note included with instruction to start a new chat each time	2
31	Clear copy boundaries — participant knows exactly where to start and stop copying	3

#	Check	Pts
32	"What this does" explanation present	2
33	Suggested email template present and customizable	2
34	"What's embedded in the prompt" section present	2
35	"What to expect back" section describes the output participants will produce	2
36	"How to review" section provides a checklist for evaluating submissions	2

These must be fixed regardless of score. A document with any blocking failure cannot be deployed.

Failure	Why It Blocks
Client's goal or objectives are wrong, incomplete, or from the golden example instead of the actual client	Participants will produce tactics aligned to the wrong objectives
Quarter dates are wrong	The entire backward plan will reference the wrong months and weeks
Output format doesn't match the client's template	Participants can't copy/paste into the template — creates friction instead of removing it
Prompt is missing one or more of the 6 steps	The conversational flow breaks — participants get incomplete output
Push-back instruction removed from Step 2	Without push-back, the LLM accepts vague tactics — defeats the purpose of the tool

Failure	Looks Like	Root Cause	Fix
Objectives from golden example left in prompt	"[Original client objective]" appears in a different client's deployment	Copy/paste error — builder forgot to replace the original deployment client's objectives	Replace all objectives with client's actual objectives
Week dates don't match quarter	Output template says "Apr 6-10" but the deployment is for Q3	Dates not recalculated for the target quarter	Recalculate all 12 weekly date ranges from the quarter start date
Golden example references removed	No example in the prompt	Builder thought the example was deployment-specific and deleted it	The HR golden example is intentionally generic — restore it unless a client-specific example is available
Consultant brief has no email template	Brief explains the tool but doesn't give the consultant a ready-to-send email	Section omitted during build	Add the email template from the golden example, customized for the client
Output format is markdown tables but client uses Word	Participant copies output but formatting breaks in Word	Tables in the prompt use markdown pipe syntax that some LLMs render and some don't	Test the prompt in the client's most likely LLM — if tables don't render cleanly, simplify to labeled sections
Prompt is too long for the LLM	Participant pastes prompt and gets an error or truncation	Some free-tier LLMs have short context windows	Test prompt length against Copilot (the most constrained common option) — if it truncates, shorten the golden example

Pass threshold: 90+

Blocking failures override score — fix all blocking failures before calculating the final score.