01 — CONTEXT: 1:1 Recap

Inputs, validation rules, content filter.

What this kit is — and the stance that never changes

This kit produces recaps for advisory engagements only — three situations:

1. TPC member 1:1 — a Tax Pro Collaborative member's monthly/quarterly call.
2. Team-and-one recap — a member plus their staff in the room (e.g., Rob + Natalie).
3. One-off advisory consult — an ad-hoc paid 1:1 that isn't AOS or a TPC group call.

It is NEVER an AOS done-for-you engagement. AOS DFY clients run a separate cascade in their own repos. This kit has no "DFY mode."

The stance invariant (always on, every sub-type) — load before composing anything

Kathryn does no between-session build work in this kit. The recap, and every downstream artifact, must never imply she's producing the member's deliverables.

• Banned framing (hard-blocks composition — file 04 + file 05 Step 6): "we'll work on this together," "I'll bring you a recommendation/build," "I'll set up / build X for you," or a Kathryn Action Item that is the member's own tooling / infrastructure / build.
• A member's build / tooling / infrastructure need is NOT a Kathryn deliverable. It routes to the signal (TPC member — a possible Sprint/offer; file 07) or a note (one-off consult). Never to a drafts/ deliverable, never to "Kathryn's Action Items," never to the agenda as Kathryn-owned.
• A build-need becomes Kathryn's work only as a paid deploy sprint / one-time build or a cross-TPC rollout — and that's a separate conversation, not this recap.

This is the Rob 6/17 failure encoded: the recap implied Kathryn would solve his SharePoint setup. She doesn't. (Memory: feedback-tpc-members-not-aos-no-between-session-builds.)

Sub-type drives SHAPE, not stance

The stance above is identical for all three. The sub-type only changes the recap's shape:

Detect the sub-type at Step 1 (from the attendees + where the member lives) and apply the matching shape. The stance never branches.

Required Inputs

Validation Rules

1. Transcript must exist. Do not draft from a relay summary or memory.
2. Reference data read first. Before transcript, before anything. Every proper noun in the email gets cross-checked. Cohort reference data is part of this — locked cadence vocabulary lives there (e.g., for TPC: Monday = Momentum Monday, Thursday = co-working). Loading cohort terminology is not optional.
3. Reference data overrides transcript. Transcripts misspell names. The reference data wins. 3b. The chat overrides the transcript for anything typed. A list Kathryn types into chat is exact; the same list read aloud and transcribed is lossy. When the two disagree, the chat wins. (Rob 8/12: the spoken pass through his categories included "out of office"; the list she actually typed had 8 and did not.)
4. Prior recap matters. If one exists, read it for voice and format match. The new recap should feel like the next email in the sequence.

Input Priority Hierarchy

When inputs conflict:

1. Reference data — wins on proper nouns, always
2. Transcript — wins on what was said, what was committed to
3. Kathryn in conversation — wins on framing, priority
4. Prior recap — wins on voice and format
5. Member file — wins on engagement context

Content Filter — What's In / What's Out

In the Email

• Everything relevant Kathryn typed into the meeting chat — reproduced, not referenced. Kathryn's rule, 2026-08-12: "he may not have grabbed them from the chat. We should never assume they do. My recaps always include relevant data from the chat." The chat panel is ephemeral to the client — they may never scroll back, and it disappears entirely if they did not save it. The recap is the durable record. If she typed a category list, a set of targets, a list of inputs, or a naming convention, it appears in full, in the email body, and in the section where the client acts on it — never as a pointer to "the list from chat."
• Wins / progress — factual, no editorializing
• Topics covered — compressed to one line each
• Action items — specific, grouped by owner (Rob's items / Kathryn's items)
• Proposed agenda for next meeting
• Warm close — short, specific to something from the call

NOT in the Email

• Backstory the client already knows (they were on the call)
• Personal venting / tax season stress (unless it's the action item)
• Coaching language, motivational framing, narration of behavior
• Stacked encouragement
• Sub-bullet recaps that mirror the transcript

What Goes in the .md (and NOT the email)

• Kathryn's takeaways
• Connections to follow up
• Ideas sparked
• Notes on the member's trajectory / patterns
• Anything candid

The .md is your private reference. The email is the client's record. Keep them separate.

Gap Protocol

If something is unclear in the transcript — ambiguous date, hedged commitment, unclear action item owner — flag it to Kathryn before drafting. Don't guess. A wrong date or misattributed action item erodes trust faster than a missing item.