agentleFS
peterbamuhigirepeterbamuhigire/business-plan-skills/skills/meta-finance/meta-investment-committee-red-team/SKILL.md

meta-investment-committee-red-team

Skill4 starsChanged 19 days agoMITView on GitHub

Use when simulating a lender, DFI, VC, grant or owner-board investment committee against a complete plan, model and evidence pack; distinguishes `meta-bankability-scoring`, which scores lender readiness without adversarial committee rounds.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Test whether a complete funding case survives audience-specific challenge.

    line 13
  2. Very specific:

    Use meta-bankability-scoring when only a lender-readiness score and fix list are required.

    line 24
  3. Very specific:

    Use meta-due-diligence when verifying records and claims rather than simulating a decision forum.

    line 25
  4. Very specific:

    Use meta-valuation when the controlling task is to calculate enterprise or equity value.

    line 26
  5. Distinctive:

    Do not run a decision round on an early concept with no reconciled model or funding ask.

    line 27
  6. Distinctive:

    Run independent challenge rounds for commercial logic, execution, finance, evidence, regulation and downside.

    line 44
  7. Distinctive:

    Every challenge cites an input location or records a precise missing artefact.

    line 51
  8. Very specific:

    Use ../../../tools/investment-committee/simulate_committee.py for deterministic manifest validation and scoring.

    line 119

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Investment Committee Red-Team

Test whether a complete funding case survives audience-specific challenge. A simulation produces decision evidence and repair priorities; it never certifies approval or predicts a real committee.

<!-- dual-compat-start -->

Use When

  • A complete plan, financial model, funding ask and evidence pack need an adversarial pre-submission review.
  • The team must rehearse lender, DFI, VC, grant-panel or owner-board questions and decision conditions.

Do Not Use When

  • Use meta-bankability-scoring when only a lender-readiness score and fix list are required.
  • Use meta-due-diligence when verifying records and claims rather than simulating a decision forum.
  • Use meta-valuation when the controlling task is to calculate enterprise or equity value.
  • Do not run a decision round on an early concept with no reconciled model or funding ask.

Required Inputs

ArtefactSource/providerRequired?Behaviour when missing
Versioned plan and funding askPlan ownerYesStop the decision verdict; return intake questions only.
Formula-audited model and assumptionsFinance ownerYesMark model integrity not assessed and block approval.
Evidence and exception registerResearch leadYesTreat unsupported load-bearing claims as blockers, not neutral items.
Audience and decision mandateSponsor or funder briefYesDo not substitute another audience's criteria.
Sector regulatory gate resultsRegulatory gate ownerConditionalBlock affected launch, cash-flow or timing conclusions.

Workflow

  1. Freeze input versions, audience, requested decision, authority and conflicts; remain read-only.
  2. Validate the pack manifest and stop if the plan, ask, model, evidence register or audience is absent.
  3. Assign audience roles from the committee framework; record each role's thesis, concern and evidence request.
  4. Run independent challenge rounds for commercial logic, execution, finance, evidence, regulation and downside.
  5. Apply audience-specific conditions and blockers before calculating a score; no aggregate score overrides a blocker.
  6. Issue one result: decline, defer, conditional-progress, or progress-for-real-review, with dissent and repairs.
  7. Recover only after the owner supplies versioned evidence, then re-run and retain the prior exception trail.

Quality Standards

  • Every challenge cites an input location or records a precise missing artefact.
  • The result separates evidence, assumptions, inferences, disagreements and matters not assessed.
  • Audience criteria affect the questions and conditions; lender, DFI, VC and grant rounds are not interchangeable.
  • Formula, balance, scenario, tax, regulatory and evidence failures remain visible and cannot be averaged away.
  • Progress-for-real-review means the pack may enter human review, never that funding is approved.

Anti-Patterns

  • Inventing a committee personality or objection. Fix: derive each challenge from the mandate and supplied evidence.
  • Letting a high weighted score erase a blocker. Fix: apply blockers before recommendation bands.
  • Using lender DSCR questions as a VC decision model. Fix: load the matching audience round.
  • Calling an absent workbook check a pass. Fix: mark model integrity not assessed and defer.
  • Rewriting the plan during an audit. Fix: stay read-only and issue owner-assigned repair actions.
  • Claiming that a simulation predicts approval. Fix: state its bounded rehearsal purpose and human-review requirement.

Outputs

ArtefactConsumerAcceptance condition
Committee decision memorandumSponsor and plan ownerStates decision, vote, blockers, conditions, dissent and next gate.
Challenge and response registerSection/model ownersEvery question has evidence, an owner, a due state and acceptance condition.
Repair sequenceDelivery leadOrders only decision-changing repairs and identifies re-review triggers.

Evidence Produced

EvidenceFormatAcceptance condition
Input version manifestJSON or tableIdentifies every reviewed artefact and formula/evidence gate state.
Audience scorecardStructured recordShows criterion score, rationale, evidence pointer and blocker status.
Dissent and exception logDecision recordPreserves minority view, unavailable checks and unresolved conditions.
<!-- dual-compat-end -->

Capability Contract

The permission boundary requires only read access to the supplied plan, model audit, evidence register and gate results, plus local execution for deterministic scoring when available. The skill defaults to read-only. It may not edit source artefacts, contact funders, submit a pack, spend funds, certify compliance, or represent a real committee decision without explicit authority from the owning role.

Degraded Mode

Without execution, apply the scorecard manually and label arithmetic unverified. Without a model, source, sector review or document, return the questions supported by available inputs and mark the affected criterion not assessed; recommendation cannot exceed defer. Never infer a pass from silence, inaccessible files or a polished narrative.

Decision Rules

ConditionActionFailure or risk avoided
Any mandatory input or audience blocker failsDecline for fatal incompatibility or defer for remediable evidenceAttractive averages conceal an unfinanceable condition
No blocker, but material conditions remainConditional-progress with owners and dated acceptance testsVague approval leaves execution risk unowned
Evidence, model, gates and audience conditions passProgress-for-real-review, retaining dissentSimulation is misrepresented as funding approval
Two roles disagree on a load-bearing assumptionPreserve both cases and request the deciding evidenceConsensus language hides unresolved risk

Worked Example

A DFI debt pack shows positive cash flow but its formula audit has not tested debt-service coverage under delay. The finance role marks model integrity unassessed, the environmental role retains the site-screen condition, and the committee returns defer with two acceptance tests. A high market score does not change that result.

References

  • Five-minute reader test (Abrams, R. M. (1993) The Successful Business Plan, The Oasis Press): before role rounds, have one reviewer read only the executive summary, financials, management and exit or repayment sections for five minutes and answer: is the idea sound, is the market sufficient, are projections realistic, is management capable, and how does the funder get its money back? Any unclear answer, and the weakest-looking section, becomes the first line of challenge.
  • Committee review framework - load roles, rounds and audience-specific questions.
  • Committee scorecard - load criteria, blocker precedence and recommendation bands.
  • Worked simulations - use to calibrate evidence-linked challenges, not copy conclusions.
  • Use ../../../tools/investment-committee/simulate_committee.py for deterministic manifest validation and scoring.

Read Next

  • meta-bankability-scoring - when lender-readiness repair, rather than committee rehearsal, owns the next decision.
  • meta-due-diligence - when a challenged claim must be verified against records.
  • meta-financial-stress-test - when the committee requests a new downside or covenant case.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.