rdfitted/skills/bp-iterate

/bp-iterate

Iterate the Fitted Business Plan(s). Manages the internal canonical and the external partner/investor variant, snapshot-on-version-bump lineage, redaction enforcement between variants, and cross-document coupling.

When this runs

• User says /bp-iterate, "iterate the BP," "bump the BP," "update the business plan," "version up the BP," "create / update / refresh the external variant"
• A material trigger fires per the BP's own Iteration Log (first 2 new closes / funding closure / new client / new partner / new risk / research dossier delivered / partner-share request)

Canonical file locations

| Variant | Current canonical | Snapshot pattern | |---|---|---| | Internal (full sensitive material, founder-primary) | D:\Document Writing\Business Plans\fitted-business-plan-internal.md | D:\Document Writing\Business Plans\fitted-business-plan-internal-v0.X.md | | External (partners / investors — highest redaction tier) | D:\Document Writing\Business Plans\fitted-business-plan-external.md | D:\Document Writing\Business Plans\fitted-business-plan-external-v0.X.md |

Coupled documents (re-check on every material iteration):

• ~/.claude/strategy/fitted-expansion.md — strategy / relationship-index layer
• ~/.ai-docs/wiki/practices/offer-architecture.md — locked SKU taxonomy
• ~/.ai-docs/wiki/partners/tom-tinervin.md — Bridge C1 / EGI thread state
• ~/.claude/strategy/atlanta-ga-construction-icp.md — GA market research dossier

Versioning protocol

Core rule: versions track content state. Variants share version numbers.

A version (e.g., "v0.2") is a specific content state of the Business Plan. At any given version, multiple variants can exist — internal (full sensitive material), external (Tier B redacted), and future audience-specific tiers if added. All variants at the same version reflect the same underlying content; they differ only in what's redacted.

v0.2  ──┬── fitted-business-plan-internal-v0.2.md   (full sensitive)
        └── fitted-business-plan-external-v0.2.md   (Tier B redacted)

v0.3  ──┬── fitted-business-plan-internal-v0.3.md   (when content changes)
        └── fitted-business-plan-external-v0.3.md   (only if material for external; external can stay at v0.2 otherwise)

When external lags internal: if a v0.3 internal change is internal-only (e.g., a sensitive INTERNAL ONLY note, founder-discretion content), external stays at v0.2 — that's fine and expected. External will catch up at the next version where material changes propagate.

Administrative changes do NOT bump versions. File renames, structural reshuffles, skill codification, frontmatter housekeeping, link fixups — these fold into the current version in place. Only substantive content changes warrant a version bump.

Step-by-step protocol

1. Determine: is this an administrative pass or a content change?

   • Administrative (rename, reshuffle, link fix, skill reference update) → edit in place at current version. Update iteration log row to note the admin update. No snapshot.
   • Content change (new fact, new section, new SKU, repricing, risk surfaced, etc.) → version bump required.

2. For content changes — snapshot the current canonical(s) BEFORE editing. cp <current>.md <current>-v0.X.md for every variant being touched. The snapshot is the frozen historical record of the outgoing version; the canonical will move to the new version.

3. Edit canonical in place. Bump version frontmatter to next number. Edit content sections per the interview output.

4. Update the in-file Iteration Log with: date, version, change narrative (1–3 sentences), trigger.

5. Propagate to sibling variants if material. If internal changes affect external:

   • Snapshot external at its current version
   • Edit external with redactions applied
   • Bump external's version to match internal's new version
   • Update external's iteration log

   If NOT material for external, leave external at its current version unchanged. Don't fake-bump it just to keep numbers aligned. Version pairing is preferred when both files reflect the same content state; version skew is acceptable when external doesn't need to move.

6. Update coupled docs if warranted. Strategy doc, wiki pages.

7. Print a 5-bullet conversation summary of what changed and why.

Redaction Rules — internal → external

Apply ALL of these when generating or updating the external variant.

Client name redaction (Tier B — industry-descriptor anonymization, locked 2026-05-22)

| Internal name | External rendering | |---|---| | TOG / The Organic Gardener NYC | "a ~$3.5M NYC commercial design-build operator (rooftop landscape vertical)" | | UBS / Ultra Building Solutions | "a ~$8–9M-run-rate residential general contractor in the Southwest, expanding into commercial" | | GVB / GVB BioPharma | "a multi-region cannabinoid / specialty manufacturer (US / UK / NL operations)" | | Scatter Ashes Maui | DROP — no longer a client | | TOTUM | "an early-stage veteran-healthcare venture (partnership / equity arrangement, not paid retainer)" | | Nuveldy's / George Vallone | "a recent ~$4M-revenue prospect conversation" (drop names; preserve the $30K/mo data point) | | Front Row / Summit RB / Gerardo / Aaron Warshaw | "active SMB-construction-adjacent prospects" | | EGI / Edge Global Indonesia / Tom Jensen | "an international reforestation / carbon-credit platform (active partnership discussion)" | | Foreman (Sam's bot at UBS) | "the original Main-Brain reference deployment at a residential GC client" |

Partner / contact name redaction (refer by company, not by person — locked 2026-05-22)

| Internal name | External rendering | |---|---| | Tom Tinervin | "the anchor-client principal / strategic advisor at [generalized client identifier]" | | Billy Roman | "Vektor Growth, LLC (channel partnership)" — refer to company, drop the personal name unless context absolutely requires | | Neal Polshuk | "Cloud Capital Funding (capital advisor / AR factoring partner)" | | Ankush | "Boca Raton-based development partnership (offshore India team)" | | 04Systems / Minana / Joris | "04Systems (international implementation partner — engagement structure under discussion)" | | Zion Duffy | "a Fitted apprentice in active training" | | Sam Lamphere | "principal at [generalized Southwest residential GC client]" | | Ken Haines / Wrench Group | "the largest GA HVAC roll-up consolidator" | | Bidblox / Lisa Mohapatra / Katrina Sitkovits | "an AI bid-management SaaS in active channel-partnership discussion" | | Andreas / Adnan / Skipper / IHC / EHC / UHC | "Middle East / sovereign-security partners in the international reforestation ecosystem" |

Keep named externally: Ryan Duffy (founder), Vektor Growth LLC (company only), Cloud Capital Funding (company only), Fitted Automation LLC. Company names of public infrastructure providers (Anthropic, OpenAI, Google, Orgo, Hermes, AgentMail) are fine.

Financial figures (Medium precision — pitch-deck grade, locked 2026-05-22)

| Internal data point | External rendering | |---|---| | Exact ARR | Rounded ("~$220K ARR" or "$200K+ ARR") | | Exact net income | Margin band ("~70% net margin") | | Exact cash position | DROP entirely | | CC carry / interest detail | DROP entirely | | Owner draws cumulative | DROP entirely | | Owner-comp structure (W-2 vs draws) | DROP entirely | | Balance sheet specifics | DROP entirely; preserve "no long-term debt" framing if true | | A/R aging detail | DROP entirely | | 2027 ARR target | KEEP as range ("$500–800K") | | Capital ask | KEEP as range ("$200–350K") | | Pricing tier | KEEP ("$7–10K/engineer-month standard; premium tier $10–15K for accelerated delivery") | | Use-of-proceeds line items | KEEP categories ("FTE hiring bridge," "Curriculum development," "Atlanta footprint," "Local-compute capex," "Working capital reserve"), GENERALIZE exact dollars to share-of-total percentages or rounded | | Nuveldy's $30K provenance | "Recent prospect conversation validated $30K+/mo willingness at the $4M-revenue tier — directional signal pending second confirmation" |

Section-level redactions / softenings

| Internal section | External treatment | |---|---| | "INTERNAL ONLY" inline blocks anywhere | DROP entirely | | Section 3.1 Owner draws / W-2 structure | DROP entirely | | Section 4.4 Moat — "Main Brain is a pattern, not IP" + "Compound Engineering is open pattern" | SOFTEN — "Fitted's moat thesis centers on institutional capability built via training pipeline, not product-level IP. Methodology is built on open patterns (Compound Engineering, MCP, agentic engineering) integrated into a battle-tested operating system." | | Section 6.6 "Negotiate from strength" — full PE-roll-up references including Wrench Group / Ken Haines | KEEP the canonical opener framing; generalize the named consolidators to "active PE consolidators in the vertical" | | Section 7.3 Bridge C1 — TOG-specific mechanics, C1 vs C2 negotiating posture | GENERALIZE to "anchor-client license-and-integration carve-out in active discussion (sized $1–2M one-time + per-acquisition integration fees)" — drop TOG name, drop C1 vs C2 framing | | Risk #8 "intentionally blank" placeholder | DROP entirely — no visible slot, no parenthetical, nothing | | Appendix A — Redaction Map | DROP entirely (it's meta about variant-generation, not for partner audience) | | Appendix B — Risk-8 placeholder | DROP entirely |

Tone adjustments

• Internal: opinionated, decision-forcing, brutal honesty, founder-talking-to-himself voice
• External: pitch-forward, investor-credible, honest but not raw

Examples:

| Internal phrasing | External phrasing | |---|---| | "Today's moat is Ryan personally." | "Fitted's current capability is founder-centric. The Business Plan codifies the conversion to institutional capability via training pipeline." | | "Main Brain is a pattern, not defensible IP." | "Main Brain is Fitted's productized integration layer — built on open patterns, hardened across multiple client engagements, voice-first by design." | | "Ryan is undercompensating himself at $40K W-2." | DROP — not appropriate for external | | "The unstated risk (h) intentionally blank." | DROP — not visible externally | | "If TOG goes sideways, 30% of revenue gone in 30 days." | "Three-way client concentration today; primary near-term focus is diversification velocity through the channel partnership and Atlanta market entry." |

Iteration triggers

Version bumps fire on any of these:

• v0.X+1 — pricing-power validation. First 2 new engagements close at confirmed tier (or push back at lower tier).
• v0.X+1 — funding stack closure event. Bridge C1 closes / SBA approved / AR factoring activated.
• v0.X+1 — material new fact. Research dossier delivered, new client signed, new partner activated, new risk surfaced, new SKU formalized.
• v0.X+1 — partner-share refresh. A specific reader wants a custom variant (e.g., a Tom-specific carve-out beyond standard external).
• v1.0 — first external-audience pitch ready. External doc baseline stabilized; can be shared without further customization.

Each version bump:

1. Snapshots prior canonical
2. Edits canonical in place
3. Updates Iteration Log
4. Propagates to sibling variant if material

Interview Protocol — Tiered

Every invocation runs an interview proportional to the size of the change. Detect the tier first, then run only that tier's interview — don't over-interview on small updates, don't under-interview on big ones.

Tier detection (run this first)

Ask the user one open question: "What changed?" Take the answer and route:

• Tier 1 — LIGHT. Typo fix, single fact correction, link addition, clarification, version-log housekeeping. "Fix the typo in section 6.3" / "Update Billy's commission rate from 20% to 22%" / "Add a link to the Atlanta dossier."
• Tier 2 — MEDIUM (default). New client closed, new partner activated, pricing data point, new risk identified, milestone reached, college partnership signed, new SKU formalized, research dossier delivered. Most iterations land here.
• Tier 3 — MAJOR. New anchored vertical, business-model shift, founder location change beyond Atlanta, ARR target band revision, ICP redefinition, capital ask redefined, new offer-architecture taxonomy, exit-strategy change.

User override escape hatches:

• /bp-iterate quick or --quick flag → force Tier 1 (skip interview, just ask "what + where + log narrative")
• /bp-iterate full or --full flag → force Tier 3 (full multi-dimension grill)
• Default with no flag → detected tier (Tier 2 most common)

Tier 1 — LIGHT interview (3 questions)

1. What's the change? (1–2 sentences; user already answered this in the detection step — confirm)
2. Which section(s) does this affect? Section numbers preferred; "I don't know" is fine and skill should infer.
3. Material enough to propagate to external? Y / N. Default N for typos and small clarifications; Y if any user-facing fact changed.
4. (closing) Iteration-log narrative — one sentence: why this change?

Tier 2 — MEDIUM interview (~8 questions, conversational, not rigid sequence)

Gather this information through natural conversation — 2–3 questions per turn, processing answers as you go. Don't fire all eight at once.

1. What's the trigger event? (the change in the world — closed a deal, met a partner, hit a milestone, etc.)
2. What are the new facts? Specific numbers, names, dates, terms. The verifiable substrate.
3. Which sections need updating? Likely candidates: Current State (Section 3), Offer / Pricing (Section 4), Target State (Section 5), The Plan (Section 6), Capital (Section 7), Risk Register (Section 8), Milestones (Section 9), Open Decisions (Section 11). Skill should propose which sections based on trigger type.
4. Does this validate or invalidate any existing assumption? E.g., does this new $7K close validate the pricing-power thesis, or push the floor higher? Does this risk's manifestation increase severity?
5. Does this resolve any Open Decision (Section 11)? If yes, mark resolved and explain how.
6. Does the moat thesis shift? Usually no for Medium, but check. (Section 4.4)
7. Material for external? Y / N. If Y, which redactions need adjustment? (E.g., new client = new anonymized industry-descriptor needed; new partner = new "refer by company" mapping needed.)
8. Coupled docs to update? Strategy doc (fitted-expansion.md), wiki partner thread, wiki offer-architecture, Atlanta dossier — any that the trigger affects.
9. (closing) Iteration-log narrative — 1–2 sentences: why this change, what it unlocks.

Tier 3 — MAJOR interview (full multi-dimension re-grill)

For structural shifts, re-grill against the dimensions that materially changed. Use the original Q1–Q6 grill structure from the BP-creation conversation (audience & ask / money truth / moat / team & build / use of proceeds / risk register) — but only the dimensions affected.

Standard dimensions to consider:

• Money truth — does the financial picture change? (run QB pull + balance-sheet check)
• Moat / defensibility — does the moat thesis shift? Run hostile-scenario test if competitive landscape changed.
• Team / capacity — does the headcount / hiring sequence / outsource structure change?
• Offer / pricing — new SKU, new tier, repricing event, taxonomy update?
• Capital — funding-stack change, capital ask revision, use-of-proceeds restructure?
• Risk register — new high-severity risk, mitigation shift, risk severity downgrade after validation?
• Geography / market — Atlanta plan shift, new ICP, vertical drift?

For each affected dimension, run the corresponding question set. Process conversationally over multiple turns.

Closing for Tier 3: in addition to the iteration-log narrative, ask: "Does this version warrant a snapshot-and-rename of the external doc to a fresh major version (e.g., external v0.X → v1.0)?" — Tier 3 changes often justify external-variant major-version bumps if the business shape has shifted.

After the interview — apply the changes

Once the interview has gathered enough information, proceed to the Workflow steps below. Don't skip — every tier still snapshots, edits canonical, propagates if material, and updates the iteration log.

---

Workflow when invoked

Step 1: Run the Interview Protocol

Per the section above. Detect tier, ask proportional questions, gather facts. Use the AskUserQuestion tool for binary/multiple-choice questions (e.g., "Material for external? Y/N"); ask conversationally for open-text.

Common trigger patterns (used for tier detection):

• "I closed a new client" → Tier 2 (pricing-power validation + new client entry + Section 5/6 update)
• "Tom's v7 closed" → Tier 2 (Bridge C1 funding-stack update + use-of-proceeds resize) — could be Tier 3 if the deal restructures Fitted's cap table
• "New research arrived" → Tier 2 (integrate into thesis / market / risk sections)
• "Need to share with X" → Tier 1 if just a refresh, Tier 2 if new redactions needed
• "We're pivoting to <new vertical>" → Tier 3 (re-grill)
• "I want to bump the pricing tier" → Tier 2 (Section 4.3 + Section 5 + Risk #3 update)

Step 2: Determine scope

• Internal-only change? (e.g., founder-only insight, internal-comp shift) — edit internal only, no propagation
• Material change affecting external? — edit internal, then propagate with redactions
• External-only request? — refuse; route through internal first. External never leads internal.

Step 3: Snapshot

cp <current canonical>.md <current canonical>-v0.X.md for every file being touched.

Step 4: Edit canonical(s) in place

Use Edit tool. Bump version frontmatter. Update content sections per the trigger. Update Iteration Log with new row.

Step 5: Propagate redactions if needed

If editing internal AND change affects external: apply Redaction Rules above, edit external in place, bump external version, update external's Iteration Log.

Step 6: Check coupled documents

For each coupled doc, decide: does this iteration warrant updating it?

• fitted-expansion.md — update if relationship table or growth plan changed
• offer-architecture.md — update if SKU taxonomy / pricing / moat language changed
• partners/tom-tinervin.md — update if Bridge C1 / EGI thread state changed
• atlanta-ga-construction-icp.md — update if new GA market intel arrived (rare)

Step 7: Summary

Print a 5-bullet conversation summary:

1. What version is the canonical at now
2. What changed (1–2 sentences)
3. Whether external was propagated
4. Whether coupled docs were updated
5. What the next iteration trigger is (so user knows when to call /bp-iterate again)

Initial creation — when external doc doesn't exist yet

If fitted-business-plan-external.md doesn't exist, this is the first run. Steps:

1. Snapshot current fitted-business-plan-internal.md to -internal-v0.X.md
2. Create fitted-business-plan-external.md by carving the internal doc through the Redaction Rules
3. External starts at its own v0.X tracking the parent internal version (e.g., if internal is v0.2 when external is created, external = v0.1 but "based on internal v0.2")
4. Add iteration log row to external noting initial creation
5. Update fitted-expansion.md cross-references to point at both

Rules

• External never leads internal. Always propagate internal → external, never reverse. If a stakeholder requests an external change, route through internal first.
• Snapshot before EVERY edit. Even tiny ones. Lineage is the audit trail.
• Iteration Log is the narrative. Git history alone is insufficient — the in-file log captures why.
• Redaction is rules-driven, not judgment-driven. When in doubt, redact harder. The Redaction Rules table is the source of truth.
• Coupling is bidirectional but with leader rules: internal leads external; strategy doc and BP are peers (either can lead, both can propose changes).
• "INTERNAL ONLY" blocks are never visible externally. No "see internal version" placeholders either — drop them silently.
• Don't sanitize internal for an external future-audience. The internal doc's value is its candor; preserving that candor is more important than worrying about who might one day see it. Use the External variant for any external sharing.

Cross-references

• The Fitted BP itself — see canonical paths above
• Strategy doc: ~/.claude/strategy/fitted-expansion.md
• Wiki — Fitted offer architecture: ~/.ai-docs/wiki/practices/offer-architecture.md
• Wiki — partner threads: ~/.ai-docs/wiki/partners/tom-tinervin.md, ~/.ai-docs/wiki/partners/04systems.md
• Wiki — ICP: ~/.ai-docs/wiki/practices/icp.md
• Wiki — productization frames: ~/.ai-docs/wiki/practices/productization-narrative.md
• Atlanta dossier: ~/.claude/strategy/atlanta-ga-construction-icp.md
• Skill provenance: created 2026-05-22 after first external-variant carve request; codifies the redaction tiers locked in the same conversation (Tier B clients, refer-by-company partners, Medium-precision financials).
• Skill iteration log:
  • 2026-05-22 v1.0 — initial skill creation with redaction rules + versioning + workflow
  • 2026-05-22 v1.1 — added Tiered Interview Protocol (LIGHT / MEDIUM / MAJOR) with detection rules + escape-hatch flags (quick / full); workflow Step 1 now invokes the interview before any edit
  • 2026-05-22 v1.2 — versioning rule made explicit: versions track content state, variants share version numbers, administrative changes fold into the current version without bumping. Concurrently retroactively rolled back the premature v0.3 admin bump on internal BP; renumbered external from v0.1 to v0.2 to pair with internal v0.2 (its source content state).
  • 2026-05-22 v1.3 — canonical BP file location updated: all BP files (internal canonical, external canonical, all snapshots) now live under D:\Document Writing\Business Plans\ rather than the parent D:\Document Writing\. Obsolete "Initial rename" stub removed. Admin pass — no BP version bump (the BP content didn't change, only its filesystem location).