01 — CONTEXT: TPC Post-Momentum-Monday Kit

Input definitions, validation rules, and what each mode requires.

Mode 1 Inputs

Validation Rules — Mode 1

1. Transcript must exist. No transcript → no run. Do not draft from memory or a relay summary.
2. Read member files BEFORE the transcript. Every member's prior context matters for interpreting what they say or don't say. The transcript means different things depending on who's in the room.
3. Read the facilitation guide. The facilitator review is a comparison; without the guide, there's no review.
4. Identify attendees and absentees first. Cross-check the transcript's speaker list against the TPC member roster in CLAUDE.md. Anyone in the roster who doesn't speak is absent — note whether their absence has a reason (medical, prior notice) or is unexplained.
5. Member-file updates only for attendees + brief absence notes for the rest. Do not invent contributions.

Mode 2 Inputs (Improve)

Validation Rules — Mode 2

1. Compare the kept version to the drafted version. Every difference is a signal.
2. If voice changed → update the output skill's voice rules + golden example pointer.
3. If structure changed → update the output skill's templates.
4. If QC missed a problem → add the check to file 04 + Common Failure Modes.

Input Priority Hierarchy

When inputs conflict:

1. TPC reference data file — wins on all proper nouns, firm names, tools, canonical spellings, prior commitments. Always.
2. Member files (CLAUDE.md and per-member) — win on member context, watch-fors, current threads
3. Transcript — wins on what was actually said in the session
4. Kathryn in conversation — wins on framing, priority, strategic intent
5. Facilitation guide — wins on what was planned; the gap between guide and transcript is the facilitator review
6. Prior session artifacts — win on voice, format, carry-forward threads
7. Planning doc — wins on arc context

Cadence Rule (Hard Constraint)

Members get exactly 2 emails per week:

• Monday recap (Output 1) — sent same-day after Momentum Monday
• Friday review and preview (Output 8 → produced as skeleton; Kathryn fills and sends Friday)

The kit does NOT produce member-facing emails outside this cadence. No personal reach-out emails to absentees. No mid-week prompts. No "just checking in" follow-ups. If a special reason arises (a member is in a crisis, a logistical issue needs urgent comms), Kathryn handles it manually — the kit does not auto-draft it.

Member-touch action items are explicitly OUT OF SCOPE for Output 10. Personal reach-outs to absent members are Kathryn's judgment call, not a kit-produced artifact. The action items the kit produces are limited to:

• Build/listen items — deep-work blocks (listen to recordings, build kit scaffolds, draft agendas)
• Decision items — things to decide before the next Monday call
• Doc-update items — kit-internal updates (member files, planning doc, lessons-learned, reference data file)

This rule is a hard constraint. A run that produces a member reach-out email outside the Monday/Friday cadence has failed QC.

Content Filtering: What Goes Where

This kit transforms one transcript into eight different artifacts with three different audiences. Filtering by audience is non-negotiable.

Monday recap email (member-facing)

In:

• Pattern-level summary of what was discussed
• Member contributions named at the level the member would be comfortable seeing in writing
• Direct asks (deadlines, deliverables)
• Housekeeping (subscriptions, scope, absent-member notes)

Out:

• Anything individual that could feel awkward called out (per /recap-meeting rules)
• Kathryn's facilitator-internal observations
• Coaching language about a member's behavior
• Quotes that would embarrass anyone

Personal facilitator recap (Kathryn-facing)

In:

• Everything from the room
• Kathryn's takeaways linking to her business / clients / IP
• Quotes worth remembering
• Member-by-member contributions in detail
• Action items
• Ideas sparked

Out:

• (Almost nothing — this is the comprehensive internal reference)

Facilitator review (Kathryn-only, your-eyes-only)

In:

• Honest comparison to the facilitation guide
• What worked, what didn't, where Kathryn deviated
• Airtime / dominance patterns
• Calibration notes for next session
• Self-questions Kathryn should sit with

Out:

• Anything Kathryn would write differently if a co-facilitator were reading
• Sanitized framing of difficult observations

Member files (Kathryn-facing, per-member)

In:

• New contributions from this session
• Updates to engagement profile if it shifted
• New watch-fors that fired in the session
• Updated topic candidates for the three events
• Commitments the member made

Out:

• Cross-member observations (those go in the personal recap or facilitator review)
• Anything not specific to this member

Planning doc append (Kathryn-facing)

In:

• Arc-level updates: what changed about the cycle's shape
• New artifacts produced this week
• Threads that surfaced and where they get carried
• Facilitator-review insights at the arc-shaping level

Out:

• Member-specific detail (those go in member files)
• Single-session takeaways unless they shift the arc

Gap Protocol

A gap means: data needed to produce an output that isn't in the transcript or member files.

Common gaps:

• A member referenced a tool, person, or process not in their member file
• A commitment was implied but not confirmed
• A date was discussed but ambiguous
• A member's absence is unexplained

The rule: Flag the gap to Kathryn. Don't guess. Don't assume.

Pre-Build Validation Gate (Gate 1 — binary)

Before starting any output, confirm:

• [ ] Transcript exists and is readable
• [ ] All member files have been read
• [ ] Facilitation guide for this session exists and has been read
• [ ] Planning doc current state has been read
• [ ] Attendee + absentee list is correct (cross-checked against TPC roster)

If any are no, stop. Get the missing input or confirm with Kathryn before proceeding to Gate 2 (the per-output QC in file 04).