Archive · Tranche 1

Capital-model safety report

Eleven flagged phrases and three priority fixes: what an agent found when told to attack the project's own finance language for leakage risk.

Original title
Capital Model Safety Report — Finance Assumptions Book, Language Review
Original date
18 May 2026
Project phase
Stage 1 — Gauging interest
Purpose at the time
An adversarial language review, by an agent, of the project's own finance assumptions book and registers: where could a correctly-hedged figure be lifted out of context and read as a return promise, repayment commitment or capital-raise target?
Status at the time
Advisory report v1.0 for steering committee review. No edits were made to the source documents by the report itself.
Source provenance
internal/capital_model_safety_report.md (private working corpus; unpublished, unchanged). The document it reviews — internal/finance_assumptions_book.md — is **not** in this tranche.
Publication treatment
Substantially intact
Derived / prepared by
Claude (Fable 5) with Adrian Wedd, 18 August 2026
Prepared
2026-08-18
Human review
Adrian Wedd — publication review completed 19 August 2026
Published
2026-08-19
What was changed for publication
  • Published verbatim. The phrases it quotes as dangerous (a percentage return figure, mortgage security, a Year-15 capital return) are carried here *as the things the report told the project to remove*; each is carried with a rule-scoped claims-allow: marker in the source so the site's claim scanner keeps checking every other line.
  • The finance assumptions book that this report reviews is not published in Tranche 1. Line references into it are left as they were.

Document begins

Capital Model Safety Report

Finance Assumptions Book — Language Review Prepared: 18 May 2026. Internal only. Advisory — no edits have been made to source documents. Reviewer: Agent (Claude). Human steering committee sign-off required before any action on these findings.


1. Summary

Documents reviewedinternal/finance_assumptions_book.md (v0.1, 729 lines), internal/assumptions-register.md, internal/strategy/risk_register.md
Flagged items11
Overall risk levelMEDIUM
Basis for MEDIUM (not HIGH)The finance assumptions book is well-structured. Its preamble, confidence-label system, and the blanket caveat at line 729 (“No number in this document has been approved for use in external or public communications”) provide a strong internal-document defence. No item in the book reads as a financial offer or solicitation. The risk is primarily one of leakage — phrases and figures that are hedged correctly in isolation but that, if extracted and quoted without their surrounding context, could be read as return promises, repayment commitments, or projections. Two items in the assumptions register (AR-13, AR-22) carry elevated financial-offer risk and are flagged separately.

2. Flagged Items

#LocationQuoted textRisk typeSuggested fix
F-01finance_assumptions_book.md §5.0A, ~line 583”optional interest up to ~5%” (Castlemaine debenture, attributed to the 17 May meeting)Return language without hedge — “optional interest up to ~5%” could be read as a precedent return figure for Cygnet’s own debentures, especially by a reader who skips the attribution note. The section heading says “attribution note” but the figure appears in a table field labelled “Figure.”Retain the Castlemaine figure but (a) move it out of the “Figure” field into the “Notes” field, (b) add a bolded inline warning: “This is a figure attributed to a different co-op’s instrument in a different jurisdiction. It is NOT a return rate for any Cygnet instrument. No Cygnet return rate has been set or approved.”
F-02finance_assumptions_book.md §5.2 (CCUs), ~line 604”Realistic size $100,000–$500,000” — no ILLUSTRATIVE label, no statement that no return rate has been determinedFigure without ILLUSTRATIVE label — the size range for CCUs is drawn from briefing research and labelled ESTIMATED, which is correct. However, the section contains no statement that the return on CCUs has not been set and cannot be assumed. A reader could infer from “realistic size” that a CCU offer of this magnitude is planned.Add a sentence at the start of the Notes field: “No CCU terms, interest rate, or return have been set. Any CCU offer requires a special resolution of members, Registrar approval, and a registered Disclosure Statement before any offer is made. This size range is indicative of comparable instruments only.”
F-03finance_assumptions_book.md §5.3 (Bank debt), ~line 618”Bank Australia might lend approximately $975K–$1.05M, requiring the remaining $450K–$525K plus stamp duty from other sources”Projected capital stack presented as a scenario — the calculation is arithmetically correct and drawn from stated assumptions. But framing it as “might lend approximately $975K–$1.05M” implies a capital structure that has not been modelled, approved, or quoted. A reader who lifts this sentence has a specific debt figure and an implied equity-raise target ($450K–$525K).Prepend the sentence with: “ILLUSTRATIVE ONLY — as a hypothetical sensitivity using assumptions 1.3, 5.3, and unconfirmed figures:” and append “No lender has been approached and no lending terms have been offered.”
F-04finance_assumptions_book.md §7 (Sensitivity Notes), ~line 712”The combined capital requirement (acquisition plus renovation at upper range with contingency) could reach $3M+, depending on scope.”Capital-raise target figure without ILLUSTRATIVE label — $3M+ is a significant headline figure. Presented without a label, it could be mistaken for a planned capital raise or a model output. It is derived from a stack of MEETING-REPORTED and ESTIMATED inputs, none of which have been verified.Prepend the sentence with “ILLUSTRATIVE SENSITIVITY — using upper-bound, unverified inputs only:” and add a closing sentence: “This figure is not a capital-raise target and is not based on verified inputs.”
F-05finance_assumptions_book.md §7, ~line 716”even a 6-month bridge on $1M of debt would cost $60,000–$90,000 in interest alone”Specific cost calculation derived from unconfirmed debt quantum — the $1M debt figure is itself unconfirmed (see F-03). Presenting a precise interest-cost range from an unconfirmed principal produces a false precision that could be quoted out of context.Prefix the sentence with “ILLUSTRATIVE — using an assumed $1M bridging principal (not confirmed):“
F-06finance_assumptions_book.md §2.13 (Contingency), ~line 320”a 20% contingency would add $180K–$240K and a 30% contingency would add $270K–$360K”Calculated figures without ILLUSTRATIVE label — these figures are correctly described as “illustrative sensitivity ranges” in the same sentence, but the label appears after the figures, not before them. The document’s preamble commits to ILLUSTRATIVE labels being carried forward. The figures could be lifted without the qualifying clause.Move the word “illustrative” to the start of the sentence, or prefix a bold label: “ILLUSTRATIVE ONLY:” before the figures.
F-07finance_assumptions_book.md §4.1 (Tourism benchmarks), ~line 517”The model has to live within those numbers, not above them.”Prescriptive modelling directive without label — this is a direct quote from the feasibility briefing and appropriate guidance. However, it is stated as a positive instruction (“has to”) without flagging that no model yet exists and no model output has been approved. The risk is that it sounds like a revenue cap that has already been operationally decided.Add a parenthetical: “(i.e., no revenue model for this project may assume rates above the regional benchmark without documented justification — this is a modelling constraint, not a forecast)“
F-08assumptions-register.md AR-13, ~line 47”optional interest rate of up to around 4–5% over 10–15 years” (Castlemaine debenture) with risk note: “citing specific interest rates for a co-op debenture crosses into financial-product territory if not carefully caveated”The register itself identifies this as a financial-product risk — the register’s own confidence note says citing specific interest rates crosses into financial-product territory. This is correctly flagged in the register, but the register entry does not state what action has been taken to prevent the figure migrating to any public document.Add an explicit action item to AR-13: “TODO: Confirm that the Castlemaine 4–5% figure appears in NO public-facing document. If it has appeared, remove or re-label. Log confirmation date and reviewer.”
F-09assumptions-register.md AR-22, ~line 56”An earlier internal draft proposed paying community investors a 5% annual dividend, returning capital at Year 15, and securing the offer with a first mortgage over the property.” (marked as removed but residual-expectation risk noted)Removed return/repayment promise with known residual-expectation risk — the register correctly flags this as a RISK with the mitigation being the audit page and meeting record. However, there is no documented process for ensuring these claims do not re-enter any future draft. “Removed from public material” is not the same as a control that prevents re-entry.Add a standing control to AR-22: “Standing prohibition: the 5% dividend, Year 15 capital return, and first-mortgage security terms must not re-enter any project document (internal or public) without explicit steering committee sign-off and legal advice. Any draft that contains these terms should be flagged to the steering committee before circulation.”
F-10finance_assumptions_book.md §5.1 (Member shares), ~line 596”Member share classification as debt or equity under AASB 132 depends on whether the co-op constitution gives the board an unconditional right to refuse redemption — this is a professional advice requirement.”Implicit redemption pathway — stating that share classification depends on a redemption-refusal clause implies that redemption of member shares is a possibility being designed for. This is technically correct and appropriately hedged. However, in a document that may inform capital modelling, the phrase “unconditional right to refuse redemption” implicitly confirms that a redemption right is under consideration. If community members later see this section, they may interpret it as a promise that shares can be redeemed.No change to the technical content is needed — it is correct. Add an explicit clarifying note: “Note: no redemption right has been decided or designed. Whether member shares are redeemable at all is an open structural question requiring legal advice and a steering committee decision before it can be included in any rules or offer document.”
F-11finance_assumptions_book.md §4.3 (comparable capital raised), ~line 538–546The table of five comparable co-ops (Grong Grong, Lockington, Broomehill, George & Dragon, Old Crown) is VERIFIED and well-sourced. However, the table is introduced with “These figures are drawn from verified case studies and serve as data points for member equity expectations.” The phrase “member equity expectations” is borderline — it sounds like the table is setting an expected equity return, not an expected equity raise quantum.Ambiguous framing — “member equity expectations” could be read as return expectations rather than capital-raise size expectations.Replace “member equity expectations” with “member capital contribution comparables” (or similar) to make clear this is about how much capital was raised, not about what members will receive back.

3. Clean Items — Sections Appropriately Hedged

The following sections are well-structured and require no immediate remediation:

  • Preamble (lines 8–16): Explicitly states the document is not a model, plan, or offer. Receivership caveat is present. Stage 1 language is correct.
  • Confidence label system (lines 20–28): The four-tier label system (VERIFIED / MEETING-REPORTED / ESTIMATED / NEEDS-PROFESSIONAL-CONFIRMATION) is clear and consistently applied throughout Sections 1–3.
  • Section 1 (Acquisition — all subsections): Every figure is appropriately labelled. The absence of a current asking price is correctly noted. The receivership context is maintained.
  • Section 1A (Heritage Status): VERIFIED labels are accurate and the source (Heritage Tasmania datasheet, Place ID 3472) is named. This section resolves the AR-01 ambiguity from the assumptions register.
  • Section 2 (Renovation): All figures are MEETING-REPORTED or NEEDS-PROFESSIONAL-CONFIRMATION. The caveat at the section heading is clear. No figure is presented as a reliable estimate.
  • Section 3 (Operating Costs): The distinction between VERIFIED legal requirements (Award, payroll tax, GST threshold) and ESTIMATED cost figures (insurance premiums, rates) is consistently maintained.
  • Section 5 introduction (line 574): “No capital instrument has been designed. No share offer is open. No Disclosure Statement has been drafted or approved.” — this is an exemplary Stage 1 disclaimer and should be used as a template for any future document that discusses capital instruments.
  • Section 6 (Unknowns table): The 20-item table of unpopulated assumptions is a strong model of epistemic discipline. No figure appears where one is unknown.
  • §5.6 CSF incompatibility (line 651): The closed/unavailable status of equity crowdfunding for a CNL co-operative is stated clearly and correctly.
  • assumptions-register.md overall structure: The FACT/ASSUMPTION/CHOICE/RISK/TODO categorisation is applied consistently. The cross-reference table and update protocol are well-designed.
  • risk_register.md R-001: The identification of “Fixed return/dividend/repayment/security claims” as the critical public-language risk (Severity: Critical) is the correct primary control. The recommended control (“Remove from public docs; solicitor review before any fundraising language”) is appropriate.

4. Top 3 Priority Fixes Before Any Public Use

These three items carry the highest leakage risk — the risk that a figure or phrase, extracted from its context in this internal document, could be read by a third party as a return promise, repayment commitment, or fundraising offer.


Priority 1 — The Castlemaine 4–5% interest figure (F-01 / AR-13)

The phrase “optional interest up to ~5%” is the single most dangerous piece of language in the combined document set. It is a specific return rate attached to a named capital instrument (community debenture). Even though it is attributed to a different co-op and is meeting-reported, it is precisely the kind of figure a journalist, regulator, or disappointed investor would excerpt to argue that the project was soliciting investment with return expectations. The AR-13 entry in the assumptions register already flags that “citing specific interest rates for a co-op debenture crosses into financial-product territory.” The document set is aware of the risk but has not closed it.

Action: Remove the figure from the “Figure” field in §5.0A of the finance assumptions book. Move it into the Notes field with a bolded prohibition marker. Add the standing action to AR-13 as described in F-08. This does not require steering committee sign-off — it is an internal document reformatting exercise, not a policy change.


Priority 2 — The $3M+ combined capital requirement (F-04)

The $3M+ figure in §7 is the most likely headline number to be extracted from this document. It is derived from a stack of MEETING-REPORTED and ESTIMATED inputs, none of which have been independently verified, and it is presented without an ILLUSTRATIVE label. If this document (or any derivative) is ever shared with a potential funder, community member, or journalist, $3M+ is the number they will remember. A project that is at Stage 1 (“gauging interest”) should not have an unlabelled capital-requirement headline.

Action: Apply the ILLUSTRATIVE ONLY label as described in F-04. This is a one-line edit to the assumptions book.


Priority 3 — Standing control on the removed AR-22 terms (F-09)

The three removed claims — 5% annual dividend, Year 15 capital return, first-mortgage security — are the most legally dangerous language the project has ever circulated. They are confirmed as removed, but there is no procedural barrier to prevent them re-entering a future draft. At Stage 2, when financial modelling begins in earnest, there will be pressure to include return scenarios and repayment timelines. Without an explicit standing prohibition documented in the register, a future drafter may re-introduce this language without recognising its legal risk.

Action: Add the standing prohibition note to AR-22 as described in F-09. Flag this control to the steering committee for endorsement at the next meeting. This is a governance action, not a document edit.


End of report. Version: 1.0 — for steering committee review. Prepared by agent review of v0.1 finance assumptions book and assumptions register. No edits have been made to any source document.

End of document ← Back to the archive