The One-Person Conglomerate

Create one row per business unit with its proposed job, owner workspace, evidence gap, and explicit non-goals. Preserve the synthetic labels and protected reserves. Do not activate memberships or authorize spending.

Inputs: starter/data/portfolio.csv, starter/docs/charter.md, starter/data/capacity.json

Outputs: deliverables/portfolio-intake.csv

Depends on: No prerequisites

• Exactly five distinct business units are represented.
• The intake names 8 reserved hours and USD 100 reserved cash.
• No row represents revenue, customers, or approval as established.

Reconcile every evidence-register row with the intake. For each unit choose one question that a low-cost experiment could disconfirm. Distinguish offline artifact inspection from later consented customer research.

Inputs: deliverables/portfolio-intake.csv, starter/data/evidence-register.csv, starter/docs/review-rubric.md

Outputs: deliverables/uncertainty-register.csv

Depends on: portfolio-intake

• All ten assumption rows are accounted for.
• Every unit has a disconfirming observation and a stop condition.
• No synthetic observation is promoted to external validation.

Inspect the original csv_guard reference and fixture. Propose a narrow local-import quality use case, an explainable finding format, and exclusions. Run only the authored offline checks; treat them as tool evidence, not market evidence.

Inputs: deliverables/uncertainty-register.csv, starter/docs/business-briefs.md, starter/reference/csv_guard.py, starter/reference/test_csv_guard.py, starter/data/csv-intake.csv

Outputs: deliverables/ledgerleaf-probe.md

Depends on: uncertainty-audit

• The duplicate invoice key in the sample is identified.
• The brief states that preflight does not certify accounting, execute formulas, or send files.
• A usable tool boundary and an unvalidated offer are separately described.

Develop an original one-page guide outline for organizing a shared supply drawer. Define a completion rubric using a fictional drawer inventory; avoid safety or equipment-maintenance instructions.

Inputs: deliverables/uncertainty-register.csv, starter/docs/business-briefs.md, starter/data/portfolio.csv

Outputs: deliverables/fieldnote-probe.md

Depends on: uncertainty-audit

• The outline includes inventory, grouping, labels, and a reversal step.
• A five-minute comprehension check is specified, not claimed performed.
• The proposed experiment maps to fn-outline or fn-observation.

Draft the field order for a purchasing-request template with plain-language labels, keyboard traversal, and a print fallback. Use only fictional request data.

Inputs: deliverables/uncertainty-register.csv, starter/docs/business-briefs.md, starter/data/portfolio.csv

Outputs: deliverables/quietbench-probe.md

Depends on: uncertainty-audit

• Required and optional fields are distinguished without color alone.
• A keyboard-only walkthrough and printable reading order are reviewable.
• No accessibility certification or customer adoption is claimed.

Describe a printable worksheet that groups fictional errands by category and available time, not navigation. Include a no-personal-address policy and a criterion for abandoning the concept.

Inputs: deliverables/uncertainty-register.csv, starter/docs/business-briefs.md, starter/data/portfolio.csv

Outputs: deliverables/routepaper-probe.md

Depends on: uncertainty-audit

• The worksheet uses categories rather than real addresses or travel predictions.
• The proposed test maps to rp-sheet or rp-interview.
• The business remains distinct from the file-tool and lesson units.

Outline a six-minute offline lesson on naming and grouping sample files. Include three original example filenames and a scored transfer exercise, with a later consent gate for learner research.

Inputs: deliverables/uncertainty-register.csv, starter/docs/business-briefs.md, starter/data/portfolio.csv

Outputs: deliverables/tinylesson-probe.md

Depends on: uncertainty-audit

• The exercise has an explicit answer rubric and no real personal files.
• Learning objectives are separate from unmeasured learning outcomes.
• The experiment maps to tl-sample or tl-transfer.

Check the ten candidate rows, money precision, selected flags, ordinal scores, and reserves. Produce a baseline reconciliation with totals and units before optimization.

Inputs: deliverables/portfolio-intake.csv, starter/data/allocation-worksheet.csv, starter/data/capacity.json, starter/docs/review-rubric.md

Outputs: deliverables/capacity-baseline.json

Depends on: portfolio-intake

• The reference selection is lf-schema, fn-outline, and rp-sheet.
• It totals 19 hours and USD 90, leaving 13 discretionary hours and USD 410.
• Scores are not converted into revenue estimates.

Run the authored allocation tests and analyze the provided worksheet. Compare the reference selection with the computed recommendation and explain deterministic tie-breaking and the one-experiment-per-unit constraint.

Inputs: deliverables/capacity-baseline.json, deliverables/ledgerleaf-probe.md, deliverables/fieldnote-probe.md, deliverables/quietbench-probe.md, deliverables/routepaper-probe.md, deliverables/tinylesson-probe.md, starter/data/allocation-worksheet.csv, starter/data/capacity.json, starter/reference/allocation.py, starter/reference/test_allocation.py

Outputs: deliverables/allocation-analysis.json

Depends on: capacity-baseline, ledgerleaf-intake, fieldnote-intake, quietbench-intake, routepaper-intake, tinylesson-intake

• A successful test summary and exact input paths are recorded.
• The recommendation respects all three capacity limits.
• The analysis reports unspent cash and unused hours rather than treating utilization as the objective.

Produce a complete ten-row proposed worksheet with selected flags and a rationale column. Accept or override the mathematical recommendation explicitly; if estimates change, retain the original values alongside proposed values.

Inputs: deliverables/allocation-analysis.json, deliverables/uncertainty-register.csv, starter/data/allocation-worksheet.csv, starter/docs/decision-template.json

Outputs: deliverables/allocation-proposal.csv

Depends on: allocation-analysis, uncertainty-audit

• The proposal contains all ten stable experiment IDs.
• At most one experiment per unit and three active units are selected.
• Any override includes a reason and preserves the protected reserves.

Write one modest offer hypothesis per business unit, with what exists now, what is proposed, and what would be learned next. Draft only: no mailing lists, outreach, publishing, or spending.

Inputs: deliverables/ledgerleaf-probe.md, deliverables/fieldnote-probe.md, deliverables/quietbench-probe.md, deliverables/routepaper-probe.md, deliverables/tinylesson-probe.md, deliverables/uncertainty-register.csv

Outputs: deliverables/portfolio-messages.md

Depends on: ledgerleaf-intake, fieldnote-intake, quietbench-intake, routepaper-intake, tinylesson-intake

• Each unit has a distinct audience hypothesis and no fabricated testimonial.
• The CSV tool is described as a reference example, not a validated commercial service.
• Every external invitation remains owner-approval-gated.

Use the decision example as a structural guide, not a prior approval. Name active and parked units, experiment exit rules, capacity totals, risks, and the exact human decision still needed.

Inputs: deliverables/allocation-proposal.csv, deliverables/portfolio-messages.md, starter/docs/decision-template.json, starter/docs/charter.md

Outputs: deliverables/portfolio-decision.json

Depends on: allocation-proposal, portfolio-messaging

• Every unit is either active, parked, or rejected with a reason.
• The decision distinguishes recommendation from owner authorization.
• There are no claimed sales, real customers, or fabricated completed experiments.

Repeat the CSV checks, document the sample's expected nonzero exit, and assemble a release-readiness checklist for the one real small reference tool. Publication is a later, separately approved operation.

Inputs: deliverables/portfolio-decision.json, deliverables/ledgerleaf-probe.md, starter/reference/csv_guard.py, starter/reference/test_csv_guard.py, starter/data/csv-intake.csv, starter/docs/review-rubric.md

Outputs: deliverables/csv-tool-handoff.md

Depends on: portfolio-decision

• The tool reads only the explicitly selected UTF-8 file and writes JSON to stdout.
• All authored tests pass; the intentionally duplicated sample key yields a duplicate-key finding.
• Distribution, signing, billing, and real-file ingestion are not represented as completed.