elliot-stack 1.0.29 → 1.0.33

package/skills/estack-leadership-coach/adding-references.md

@@ -0,0 +1,280 @@
+# How to add a reference file to the knowledge vault
+
+<primary_outcome>
+A new file in `references/` that contains live-fetched, properly cited source material — plus every placeholder link and "Real-world case" block in the skill that points to it has been updated to use the verified content from that file. After this task is complete, the user can ask the coach about that source mid-session and get a grounded answer with a real URL.
+</primary_outcome>
+
+This file is the playbook for populating the knowledge vault. The user triggers it by saying something like *"I want to add a reference source"* or *"Let's build the reference for [author/work]."* When that happens, load this file and follow it step by step.
+
+---
+
+## Hard rules (apply throughout — never violated)
+
+1. **Live-fetch every fact.** Use `WebSearch`, `WebFetch`, or `mcp__claude_ai_Supadata__supadata_transcript` (for YouTube, Instagram, TikTok, and all other social media) every single session you build a reference. Never recall content from training memory and present it as sourced. This rule comes from the user's global CLAUDE.md and is not negotiable.
+2. **Cite real URLs only.** Every reference file ends with a Sources section that contains URLs fetched this session. A name without a URL is a fabricated citation — do not write one.
+3. **Do not paraphrase quotes.** If you put text in quote marks, it must be a verbatim extract from the fetched source. Otherwise drop the quotes and frame as synthesis.
+4. **Note the fetch date.** Every reference file's frontmatter records the date the material was fetched. Source content drifts; the fetch date is how the user knows how fresh the snapshot is.
+5. **If you can't fetch it, say so.** Books that aren't online verbatim, paywalled articles, deleted videos — these can't be sourced live. Tell the user, and propose pulling from interviews / talks / podcast appearances by the same author that *are* accessible OR ask the user to provide the source. Do not fabricate to fill the gap.
+
+---
+
+## Workflow (six steps — run in order)
+
+### Step 1 — Confirm scope with the user (ask 1–3 questions, then stop)
+
+Before fetching anything, get explicit on:
+
+1. **Which source?** Specific book, article, talk, video, podcast episode — with title and author.
+2. **Which type of reference?**
+   - **Extraction** — verbatim quotes/passages organized for retrieval (best for talks, interviews, articles, video transcripts)
+   - **Synthesis** — organized key takeaways drawn from longer material (best for books, multi-source bodies of work)
+3. **Which phases / placeholders should this reference feed?** (Optional but helpful — narrows the synthesis.) For example: *"This Grove reference should feed the Phase 2 TRM case placeholder and the Phase 5 midpoint-review placeholder."*
+
+Stop and wait for the answers. Do not start fetching yet.
+
+### Step 2 — Live-fetch the source material
+
+Pick the right tool for the source type:
+
+| Source type | Tool | Notes |
+|---|---|---|
+| YouTube video / talk | `mcp__claude_ai_Supadata__supadata_transcript` | Returns the full transcript. Capture the URL and the video ID. |
+| Article on a public site | `WebFetch` or `mcp__claude_ai_Supadata__supadata_scrape` | Fetch the page and pull the prose. |
+| Web search to locate authoritative sources | `WebSearch` | Use when the user names a concept but not a specific URL. |
+| Podcast episode | Check if the podcast has a transcript page; fetch that with `WebFetch`. Otherwise look for show notes / quoted passages from secondary coverage. | Note: audio-only without transcript ≠ fetchable. |
+| Book | Look for: official author talks on the book's themes, interviews with the author, publisher excerpts, the author's own essays summarizing the book. Cite each fetched piece. | Do not fabricate page numbers or "quotes from the book" without a verifiable source. |
+
+If the first fetch is thin, do additional fetches. A good reference file synthesizes 2–4 sources, not 1.
+
+### Step 3 — Decide the type and pick the template
+
+Pick **extraction** or **synthesis** based on what you fetched and what the user asked for. Both templates are below — use the matching one.
+
+If the source is a single article or talk → extraction.
+If the source is a book or a body of work → synthesis.
+If both apply (e.g., a book with multiple author talks) → synthesis as the primary structure, with key extractions embedded.
+
+### Step 4 — Create the reference file
+
+**Location:** `~/.claude/skills/leadership-coach/references/<filename>.md`
+
+**Filename convention:** `<author-lastname>_<work-shortname>.md` — lowercase, hyphens not underscores within name parts, single underscore between author and work. Examples already in the skill:
+
+- `grove_high-output-management.md`
+- `hormozi-leila_4-stages.md`
+- `oncken-wass_monkeys-hbr-1974.md`
+
+Match the existing filenames exactly when you're populating a placeholder — the link path in the placeholder is the contract.
+
+Create the file using the appropriate template from the **Templates** section below.
+
+### Step 5 — Wire it up across the skill
+
+A new reference file is only useful if the existing placeholders find it. After creating the file, do a complete sweep:
+
+1. **Search every phase file** for the filename you just created. Each match is either:
+   - A "Going deeper" link block at the bottom of the phase → no change needed, the link already points correctly
+   - A "Real-world case" placeholder pointing to this reference → **replace the placeholder block with verified case material from the new reference file**
+2. **Search `SKILL.md`** for any inline mention of the author / work that the new reference covers. If a paraphrased claim or quote in `SKILL.md` is now backed by your verified source, update it to match the source's actual wording. If it doesn't match, hedge it or remove it — do not bend the source to fit the existing prose.
+3. **Search flow files** (`flows/pre-delegation.md`, `flows/post-mortem.md`). Less common, but check.
+
+For each "Real-world case" placeholder you replace:
+
+- Pull 1–2 paragraphs from the reference file's body — verbatim where you can, synthesized where you must
+- Keep the section header as `## Real-world case: <descriptive title>` (drop the `(placeholder — fill in during reference build)` marker)
+- Include the source URL inline or as a footnote: *"From Grove, *High Output Management*, p. 142 — see [reference](../../../references/grove_high-output-management.md)"*
+- Do **not** add invented specifics. If the reference has no verbatim case with the dialogue/metrics that would make the section pop, write a short principle illustration grounded in what *is* in the reference, and accept that it's less dramatic than the placeholder hoped for.
+
+### Step 6 — Verify (acceptance self-audit)
+
+Before declaring done, confirm:
+
+- [ ] Reference file exists at the expected path with the expected filename
+- [ ] Frontmatter includes `name`, `author`, `work`, `type` (extraction/synthesis), `last_fetched` date, and `sources` (URLs)
+- [ ] Every fact, quote, and statistic in the file is traceable to a URL in the Sources section
+- [ ] Every "Real-world case" placeholder that points to this reference has been replaced with verified content OR explicitly left as a placeholder with a note explaining why (e.g., "verified content available but not yet drafted")
+- [ ] Every "Going deeper" link block in phase files that references this file still resolves correctly
+- [ ] No hedged or fabricated content has been smuggled in — if the source doesn't say it, the reference file doesn't say it
+- [ ] Filename in the file matches the link paths used by placeholders
+
+Report back to the user with: (a) which reference was built, (b) which placeholders / cross-references were updated, (c) any placeholders that were *not* updated and why, (d) any source material the user might want to add later to fill gaps.
+
+---
+
+## Templates
+
+### Template A — Extraction reference
+
+Use for articles, talks, interviews, video transcripts — anything where you can pull verbatim text.
+
+```markdown
+---
+name: <author-lastname>_<work-shortname>
+title: <Full title of the work>
+author: <Author name(s)>
+work_type: <article | talk | interview | podcast | video transcript>
+type: extraction
+last_fetched: <YYYY-MM-DD>
+sources:
+  - <URL 1>
+  - <URL 2>
+---
+
+# <Author> — *<Work title>*
+
+## Overview
+
+<2–3 sentence framing of what this work covers and why it matters for leadership coaching. No fabrication — only what's actually in the fetched material.>
+
+## Why this is in the vault
+
+<1 paragraph: which phases / coaching moves draw on this work, and what specific principle it backs.>
+
+## Key extractions
+
+> "<Verbatim quote 1>"
+> — <Source location: timestamp / paragraph / page number if available>
+
+> "<Verbatim quote 2>"
+> — <Source location>
+
+(Pull 5–15 strong extractions. Each one is verbatim. Each one cites where in the source it came from.)
+
+## Notable cases / illustrations from the source
+
+<If the source contains specific case material — a story the author tells, a study they cite, a scenario they walk through — extract it here. Each one is faithful to the source.>
+
+### <Case 1 title>
+
+<Faithful retelling, with direct quotes where possible. Cite the location in the source.>
+
+## Where this is used in the skill
+
+- `phases/<file>.md` — <which placeholder / "Going deeper" block uses this>
+- `SKILL.md` — <if applicable>
+
+## Sources (live-fetched on <YYYY-MM-DD>)
+
+- [<Title of source>](<URL>)
+- [<Title of source>](<URL>)
+```
+
+### Template B — Synthesis reference
+
+Use for books, multi-source bodies of work, or any case where you're synthesizing from several fetched pieces.
+
+```markdown
+---
+name: <author-lastname>_<work-shortname>
+title: <Full title of the work or body of work>
+author: <Author name(s)>
+work_type: <book | body of work>
+type: synthesis
+last_fetched: <YYYY-MM-DD>
+sources:
+  - <URL 1>
+  - <URL 2>
+  - <URL 3>
+---
+
+# <Author> — *<Work title>*
+
+## Overview
+
+<2–3 sentence framing.>
+
+## Why this is in the vault
+
+<1 paragraph: which phases / coaching moves draw on this work.>
+
+## Synthesis — core principles
+
+### Principle 1: <name>
+
+<2–3 paragraph synthesis of the principle, drawn from the fetched sources. Use direct quotes where the source language is sharp; paraphrase where you're integrating across sources. Every claim should be defensible against the Sources section below.>
+
+### Principle 2: <name>
+
+<...>
+
+### Principle 3: <name>
+
+<...>
+
+(3–6 principles total. Don't pad.)
+
+## Verbatim extracts (when sources support them)
+
+> "<Quote>"
+> — <Source URL or title>
+
+(Include verbatim extracts only where you actually fetched the verbatim text. If you're synthesizing from interview snippets and don't have a clean quote, skip this section.)
+
+## Notable cases / illustrations
+
+<If the fetched sources contain specific cases — a story the author tells in an interview, a case study from a talk — extract them faithfully here. Each one cites where it came from.>
+
+### <Case 1 title>
+
+<Faithful retelling.>
+
+## Where this is used in the skill
+
+- `phases/<file>.md` — <which placeholder uses this>
+- `SKILL.md` — <if applicable>
+
+## Sources (live-fetched on <YYYY-MM-DD>)
+
+- [<Title>](<URL>)
+- [<Title>](<URL>)
+
+## Known gaps
+
+<Optional. If the user might later want to deepen this reference: name the gap. Example: "Did not fetch the original *High Output Management* text — only Grove talks and secondary summaries. Future pass could add direct chapter excerpts if obtainable.">
+```
+
+---
+
+## Cross-reference map (where to look when wiring up)
+
+For convenience, here's where each existing reference filename is mentioned in the skill body. When you build a new reference, update *all* the locations that point to it.
+
+| Reference file | Mentioned in |
+|---|---|
+| `grove_high-output-management.md` | `phases/1-intake.md`, `phases/2-trm-assessment.md`, `phases/4-build-brief.md`, `phases/5-monitoring.md`, `phases/7-diagnose.md` |
+| `gerber_e-myth-revisited.md` | `phases/1-intake.md`, `phases/7-diagnose.md` |
+| `hormozi-leila_4-stages.md` | `phases/2-trm-assessment.md`, `phases/7-diagnose.md` |
+| `hormozi-alex_followthrough.md` | `phases/4-build-brief.md`, `phases/7-diagnose.md` |
+| `doerr_measure-what-matters.md` | `phases/4-build-brief.md`, `phases/5-monitoring.md`, `phases/7-diagnose.md` (primary); `phases/1-intake.md`, `phases/3-enrollment.md` (secondary) |
+| `sullivan_who-not-how.md` | `phases/1-intake.md`, `phases/3-enrollment.md`, `phases/4-build-brief.md`, `phases/7-diagnose.md` |
+| `sanchez_main-street-millionaire.md` | `phases/2-trm-assessment.md`, `phases/7-diagnose.md` |
+| `ferriss_4hww.md` | `phases/1-intake.md` |
+| `oncken-wass_monkeys-hbr-1974.md` | `phases/6-reverse-delegation.md`, `phases/7-diagnose.md` |
+| `deci-ryan_self-determination-theory.md` | `phases/3-enrollment.md`, `phases/7-diagnose.md` |
+| `gallup_engagement-research.md` | `phases/3-enrollment.md` |
+| `van-edwards_cues.md` | `phases/5-monitoring.md` |
+
+If you add a reference not on this list, append it to this map after you finish the wire-up so future passes have an accurate index.
+
+---
+
+## Pre-empted shortcuts
+
+- **Don't research from memory and dress it up as sourced.** Every claim needs a URL fetched this session.
+- **Don't fabricate page numbers, timestamps, or quote locations.** If you don't know where the quote came from, omit the location citation.
+- **Don't write a "Notable case" with invented dialogue.** If the source contains the case, extract it. If it doesn't, leave the section empty or skip it.
+- **Don't skip the cross-reference sweep.** A reference file that exists but isn't wired up adds zero value to coaching.
+- **Don't bend the source to fit existing skill prose.** If the inline content in a phase file disagrees with what the source actually says, fix the phase file — not the reference.
+- **Don't build references the user didn't ask for.** This task is triggered by the user. Don't preemptively add references to "round out the vault" — that's how scope creep starts.
+
+---
+
+## When the user says "add a reference source"
+
+1. Ask the Step 1 questions. Stop and wait.
+2. Once the user answers, fetch the source material (Step 2).
+3. Confirm with the user: *"I fetched [N] sources for [author/work]. Building as a [extraction / synthesis]. Going to populate [N] placeholders in [phase files]. Sound right?"*
+4. Build the reference file (Step 4).
+5. Sweep and wire up (Step 5).
+6. Run the acceptance audit (Step 6).
+7. Report back with what changed and any gaps.

package/skills/estack-leadership-coach/frameworks/delegation/flows/post-mortem.md

@@ -0,0 +1,120 @@
+# Post-mortem flow
+
+<primary_outcome>
+The user finishes this flow with a written diagnosis: which of the five structural gaps caused the failure, the principle behind it, the specific moment in the prior handoff where the gap opened, and a concrete corrective move. The diagnosis closes with an offer to re-run the pre-delegation flow using the gap as the starting correction. If the user accepts, the pre-delegation flow runs immediately with the existing context.
+</primary_outcome>
+
+This flow runs when a delegation already happened and went wrong. The work came back broken, late, or off-target — or it didn't come back at all. The user is here for the diagnosis, not to recover the lost work.
+
+---
+
+## When to run this flow
+
+- The user describes a handoff that already happened and failed
+- The work came back below the bar, late, or in the wrong direction
+- The user finds themselves redoing work they delegated
+- The user keeps getting pulled back into something they thought they'd handed off
+- Trust between user and owner is fraying after the delegation
+
+If the user hasn't handed off the work yet, use `pre-delegation.md` instead.
+
+---
+
+## Phase sequence
+
+This flow runs only Phase 7, but Phase 7 produces the full diagnosis.
+
+| # | Phase | Output the phase must produce |
+|---|---|---|
+| 7 | `../phases/7-diagnose.md` | Named structural gap (1 of 5) + named failure mode + specific principle explaining why the gap caused the failure + one corrective move |
+
+After Phase 7, deliver the diagnosis artifact using the template below, then offer the pre-delegation re-run.
+
+---
+
+## How to run this flow without re-traumatizing the user
+
+The user often arrives frustrated, embarrassed, or angry — at themselves, at the owner, or both. Coach with that in mind:
+
+- **Lead with the system, not the person.** Almost every delegation failure is structural, not character-based. If you find yourself building a case against the owner, you've drifted. Bring it back to which of the five gaps opened.
+- **Don't pile on.** The user already knows it went wrong. They don't need a lecture on what they should have done. They need a diagnosis they can use next time.
+- **The owner is not in the room.** Be careful with claims about the owner's intent. Stick to observable behavior.
+- **It's almost never 'they're not capable.'** Wrong TRM calibration looks like incapability but is actually the user applying their general impression of the person to a task type the person hadn't done before.
+
+---
+
+## Pre-empted shortcuts
+
+- **Don't diagnose before asking the diagnostic questions.** It's tempting to pattern-match on the first sentence the user says ("they didn't do what I wanted"). Run the questions from Phase 7 — the surface story usually hides the actual gap.
+- **Don't pick the first gap that fits.** Multiple gaps often co-occur. Name the *primary* gap — the one that, if fixed, would have prevented the failure. Mention secondary gaps but don't dilute the diagnosis.
+- **Don't recommend a "have a talk with them" as the corrective move.** That's not a fix — it's a deflection. The corrective move is a structural change: a named authority level, an externalized success criterion, a check-in cadence, a written brief.
+
+---
+
+## Artifact template — Diagnosis
+
+When Phase 7 produces its output, deliver the artifact as a markdown block exactly like this:
+
+<template>
+
+```markdown
+# Delegation Post-Mortem
+
+**Situation:** <one-sentence description of what was delegated and to whom>
+**What went wrong:** <one-sentence description of the failure, in observable terms>
+**Team mode:** <Hierarchical | Flat>
+
+---
+
+## The gap
+**<Enrollment | Authority | Context | Success criteria | Accountability diffusion (flat teams)>**
+
+<2–4 sentences naming where in the prior handoff this gap opened — the specific moment or omission. Quote the user's own words when possible.>
+
+## The principle behind it
+<1–2 sentences of theory with attribution. Why this gap reliably causes this kind of failure.>
+
+## The failure mode this maps to
+<From the failure-mode table in Phase 7 — name the row that matches.>
+
+## The corrective move
+<One concrete, structural change that, if applied next time, prevents this gap from opening. Not "talk to them" — a specific brief element, authority level, or check-in.>
+
+## Secondary gaps (if any)
+<Optional. Other gaps that also opened, but were not the primary cause. One line each.>
+```
+
+</template>
+
+---
+
+## Closing offer (always include after the diagnosis)
+
+After the diagnosis is delivered, ask the user:
+
+> *Want to run the pre-delegation flow now, starting from the corrected gap? We'll skip what you already know about the situation and focus on building the brief / authority level / check-in structure that wasn't there last time.*
+
+If the user accepts: load `pre-delegation.md` and resume at the phase that maps to the diagnosed gap, carrying forward the task / owner / timeline / team-mode already established:
+
+| Primary gap | Resume at |
+|---|---|
+| Enrollment | Phase 3 (run all four moves) |
+| Authority | Phase 4, element 5 (then continue to 5 and 6) |
+| Context | Phase 3 Move 1 → Phase 4, element 2 (then continue) |
+| Success criteria | Phase 4, element 3 (then continue to 5 and 6) |
+| Accountability diffusion | Phase 1 (owner-selection) → Phase 4, element 6 |
+
+After resuming at the gap's phase, run forward to the end of the flow. Don't skip the artifact assembly — the corrected brief is the point.
+
+If the user declines: deliver the diagnosis and stop. Do not push.
+
+---
+
+## Acceptance self-audit (run before delivering the diagnosis)
+
+- [ ] The named gap is one of the five (Enrollment / Authority / Context / Success criteria / Accountability diffusion)
+- [ ] The principle is attributed to a specific source (Grove, Hormozi, Sullivan, Oncken/Wass, etc.)
+- [ ] The failure mode named maps to a row in the Phase 7 failure-mode table
+- [ ] The corrective move is structural, not relational
+- [ ] The offer to re-run pre-delegation is present
+- [ ] The diagnosis does not blame the owner where the gap is structural

package/skills/estack-leadership-coach/frameworks/delegation/flows/pre-delegation.md

@@ -0,0 +1,138 @@
+# Pre-delegation flow
+
+<primary_outcome>
+The user finishes this flow with two things in hand: (1) a complete Delegation Brief in markdown, ready to share with the person taking the work, and (2) enrollment talking points for the sit-down conversation that happens before the brief is shared. If the user leaves without both of those, the flow is not complete.
+</primary_outcome>
+
+This flow runs when the user has not yet handed off the work and wants to set the delegation up correctly. It orchestrates Phases 1 through 6 in order, then assembles the artifact.
+
+---
+
+## When to run this flow
+
+- The user is preparing to delegate something they currently own
+- They want help deciding whether/how/to whom to delegate
+- They want to write a brief but don't know what should be in it
+- They've tried to delegate the work before and want to start over with structure
+
+If the work has already been handed off and went sideways, use `post-mortem.md` instead.
+
+---
+
+## Phase sequence
+
+Each phase has its own file in `../phases/`. Load and follow each one in order. Do not jump ahead — a phase is incomplete until it produces the output the phase declares.
+
+| # | Phase | Output the phase must produce |
+|---|---|---|
+| 1 | `phases/1-intake.md` | Named task, named owner (or owner-selection logic for flat teams), timeline; filter decision (Eliminate / Automate / Delegate / hold); resistance pattern named if present |
+| 2 | `phases/2-trm-assessment.md` | Task-Relevant Maturity for this person on this task (Low / Medium / High) + Hormozi progression stage (Investigation / Informed Progress / Informed Results / Complete Ownership) |
+| 3 | `phases/3-enrollment.md` | Enrollment talking points: the problem, why-them, the energizing question, the needs question |
+| 4 | `phases/4-build-brief.md` | The brief: What, Why, Success Looks Like, Constraints, Authority Level (1–5), Reciprocal Commitments (flat teams) |
+| 5 | `phases/5-monitoring.md` | Check-in schedule with cadence calibrated to TRM, and what each check-in will cover |
+| 6 | `phases/6-reverse-delegation.md` | A named protocol for what the owner does when they hit a roadblock — preventing monkey-transfer back to the user |
+
+After Phase 6, deliver the artifact using the template below. Do not declare the session done until the artifact is in the conversation.
+
+---
+
+## Compressed path
+
+If all four conditions are true (trusted peer or proven high-TRM teammate, low public visibility, short timeline, low cost of failure), run a four-step compressed path instead:
+
+1. Confirm the deliverable in one sentence (Phase 1 + 4 condensed)
+2. Name "why you" in one sentence (Phase 3 Move 2 only — the other three moves are skipped)
+3. Name the authority level out loud (Phase 4, element 5)
+4. Set one check-in (Phase 5)
+
+Then deliver a shortened brief with What / Why You / Authority Level / One Check-In filled in. Skip Moves 1, 3, 4 of enrollment and skip full reciprocal commitments. Mention briefly that the compressed path is being used and why.
+
+If at any point a condition turns out to be false (the timeline grew, the visibility expanded), drop back to the full flow.
+
+---
+
+## Pre-empted shortcuts
+
+- **Don't lecture all 6 phases up front.** The user will check out. Run phases one at a time.
+- **Don't fill in the brief from your assumptions.** If the user couldn't articulate Success Looks Like, do not generate it. Push the question back until they have it.
+- **Don't skip enrollment because "they're already on board."** Enrollment is not the user's belief about the owner's buy-in — it's the talking points the user will use in the actual conversation. Always produce them.
+- **Don't deliver the brief without check-ins on the calendar.** A brief without a check-in schedule is abdication waiting to happen.
+
+---
+
+## Artifact template — Delegation Brief
+
+When all six phases are complete, deliver the artifact as a markdown block exactly like this. Fill in every field with the specific content captured during the phases.
+
+<template>
+
+```markdown
+# Delegation Brief
+
+**Task:** <one-sentence deliverable from Phase 1 + Phase 4 ①>
+**Owner:** <named person from Phase 1>
+**Timeline:** <from Phase 1>
+**Team mode:** <Hierarchical | Flat — detected during the session>
+
+---
+
+## Why this matters
+<from Phase 4 ②: the actual problem being solved, who it's for, what goes wrong if it's late or off>
+
+## Why this owner
+<from Phase 3 ②: specific reason they were chosen — not "they're great">
+
+## Success looks like
+<from Phase 4 ③: concrete description of done, with the standard externalized. Excellent / Acceptable / Poor distinctions if surfaced.>
+
+## Constraints
+<from Phase 4 ④: non-negotiables — timeline, budget, stakeholders to involve, decisions they can't make alone>
+
+## Authority level
+**Level <1–5> — <Name>**
+<one-line description of what that level means in this specific situation>
+
+## Check-in schedule
+<from Phase 5: actual cadence, e.g., "Early-stage alignment check on <date>, midpoint review on <date>, final delivery on <date>". Each check-in says what it covers.>
+
+## When the owner hits a roadblock
+<from Phase 6: the named protocol — what they do, how they bring it to the user, what the user will/won't take back>
+
+## Reciprocal commitments
+<Flat teams only — from Phase 4 ⑥: what the user/team owes the owner: blockers they'll clear, stakeholders they'll handle, decisions they'll stay out of>
+```
+
+---
+
+# Enrollment talking points
+
+**To use before sharing the brief — these are for the sit-down conversation.**
+
+> **The problem we're solving:** <from Phase 3 ①>
+>
+> **Why it matters right now:** <stakes, urgency, downstream impact>
+>
+> **Why you're the right person for this:** <specific, not generic — from Phase 3 ②>
+>
+> **What part of this energizes you?** <ask in the conversation, listen to the answer — Phase 3 ③>
+>
+> **What would help you do your best work?** <ask, listen — Phase 3 ④>
+
+</template>
+
+---
+
+## Acceptance self-audit (run before declaring the session done)
+
+Before delivering the artifact, silently verify all of these. If any is false, return to the relevant phase rather than ship a half-done brief.
+
+- [ ] The deliverable is specific enough that a stranger could tell if it was met
+- [ ] Success Looks Like is concrete — not "polished" or "high-quality" or "good"
+- [ ] The authority level is named explicitly with a number and a name
+- [ ] At least one check-in is on a calendar date, not "we'll figure it out"
+- [ ] If flat team: reciprocal commitments are filled in with specific items
+- [ ] The roadblock protocol from Phase 6 is named (not "they'll come to me")
+- [ ] Enrollment talking points include a specific "why you" — not a generic compliment
+- [ ] The user has not said "change outcome" without being re-routed
+
+When all are true, deliver the artifact and then ask: *"Want to walk through how to actually open the enrollment conversation, or are you good to go?"*