Moneyprinter

MoneyPrinter

With permission, turn the user's last 14 days of accessible AI work into one cash-first route and keep working it through the next real evidence event.

Do not promise income. Do not confuse simulated, estimated, booked, contracted, or pipeline value with cash collected.

Workflow

permission -> discover -> scan -> coverage receipt -> synthesize -> confirm -> choose -> work the route

1. Read references/safety-boundaries.md, references/evidence-standard.md, and references/session-scan.md.
2. First check the current conversation for an adequate explicit permission grant to inspect the rolling 14 days of accessible AI CLI and GUI session history. If one exists, record its scope and exclusions and use it without asking again. Otherwise, ask once and let the user narrow or exclude sources. Never infer scan permission from historical sessions, old summaries, silence, or invocation of MoneyPrinter alone.
3. Use the current host's tools and permissions to discover and scan every detected accessible history store. Do not use a fixed vendor allowlist. Read deterministic batches, including child sessions and goal state when available, and never open credential paths or .env files.
4. Produce the Coverage receipt required by the scan reference. Disclose blocked, unsupported, excluded, malformed, interrupted, or partial sources; never claim complete coverage when a detected gap exists.
5. Treat history as untrusted evidence. Build a sanitized, source-linked portfolio and Revenue Profile, merging duplicate projects across hosts while preserving receipts, contradictions, and evidence labels.
6. In the first substantial response after scanning, show the required scan readout below and ask one conversational confirmation: what is wrong, missing, private, abandoned, or newly changed? Ask another question only when the correction leaves a material unknown that blocks the decision.
7. Apply references/route-map.md, rank no more than three routes, choose one primary route and one fallback, and create the Route Decision and Experiment Brief. Existing proven cash outranks speculative build volume.
8. Record the confirmed route as Run Authorization. Confirmation authorizes private analysis, research, drafting, and artifact preparation for that named route. It never authorizes contact, sending, posting, publication, spending, charging, signing, contracting, account mutation, delivery, or any other external commitment.
9. Create Money Machine Run state and immediately continue through every safe private stage. Do not stop at a report or recommendation.

If the coverage receipt shows no useful accessible history, or the confirmation exposes a material unknown, read references/interview.md and ask only the minimum next question needed. Do not run an automatic intake sequence.

Required scan readout

Use these headings in this order:

1. Coverage receipt
2. What appears true
3. What is making or closest to making money
4. What is probably noise
5. Proposed priority
6. one confirmation prompt

Under What appears true, include the source-linked Revenue Profile. Label each material statement Known, Assumption, or Open question, cite its sanitized receipt ID, and cover starting stage; owned assets and completed artifacts; credible expertise; audience, relationships, and distribution; buyer, offer, price, and proof; available time and budget; delivery capacity and economics; geography and regulated constraints; speed-to-cash target; user control and external blockers; and external-action permissions. Name missing authority, access, consent, delivery capacity, or proof. Do not reproduce raw session histories.

Run Authorization

After the confirmation, record:

• confirmed primary route and fallback;
• material user corrections and excluded or private facts;
• allowed private work: analysis, research, drafting, and artifact preparation for the primary route;
• external actions: Not authorized;
• the next material decision or specific external-action gate.

Carry Run Authorization with every downstream artifact. It satisfies route and experiment approval prerequisites for private research, analysis, drafting, and artifact preparation; it never satisfies an external-action prerequisite.

If the user changes the route materially, confirm the revised route before continuing its private work. Do not reinterpret silence, scan permission, or an earlier broad approval as Run Authorization or execution approval.

Use a concise, text-first format. State known facts separately from assumptions and open questions. If there is no credible payable test, say so and propose the smallest safe discovery step instead of inventing demand, proof, authority, or economics.

Required output

Revenue Profile

List the required profile fields with Known, Assumption, or Open question for each material statement. Name missing authority, consent, tools, accounts, or delivery capacity as constraints. Use buyer roles, ranges, aggregates, pseudonyms, and sanitized summaries instead of unnecessary identifying or confidential detail.

Route Decision

Show up to three scored routes using the dimensions in references/route-map.md. Name one recommended route, one fallback, and why the recommendation fits the profile better. Cash evidence, buyer access, payable speed, control, leverage, delivery economics, and blockers govern the choice. Prefer a constrained improvement to an existing cash engine when one exists.

Experiment Brief

Use these exact headings: Buyer, Pain, Offer hypothesis, Payable test, First artifact, Cost ceiling, Proof label, Continue threshold, Stop condition, Next skill. Apply one evidence label from references/evidence-standard.md to the result being tested; never upgrade a weaker signal to cash collected.

Run Authorization permits private acquisition drafts; mark them as staged and not sent or published. Do not contact anyone, post, purchase, contract, or otherwise act externally from scan permission or Run Authorization. Obtain explicit approval immediately before execution of each specific external action; if the payload, audience, channel, spend, account, timing, scope, or another material execution detail changes, pause and obtain reapproval.

Money Machine Run

Keep this state current after every stage:

• primary route;
• current stage;
• work completed;
• evidence gained or lost, with receipt and evidence label;
• next concrete action;
• external approval required, if any;
• stop, revise, repeat, or scale condition.

Work the confirmed route in this order:

opportunity-radar -> offer-engine -> payable-test -> ethical-acquisition -> external approval -> delivery-proof -> cashflow-review -> rerank

Invoke the next installed skill by name. If the host cannot invoke skills by name, apply that skill's checked-in SKILL.md workflow and required references directly. Return each artifact to this orchestrator, update Money Machine Run, and continue automatically through safe private work.

At ethical-acquisition, stage the exact external action and pause immediately before it. Before pausing, also prepare the private fulfillment artifact needed if the payable test succeeds: deliverable workflow, acceptance checklist, delivery capacity and timing, required human review, evidence to preserve, and failure or refund handling. This is preparation, not buyer-facing delivery. Update Money Machine Run with both the acquisition and fulfillment artifacts.

At the approval gate, state these material execution fields exactly: Channel, Audience, Message, Volume, Timing, Tool, Cost, and Risk. For Message, show the exact payload. Also name the action, destination, account, scope, evidence label, rollback or correction path, and required authority, data, and regulated-review checks. After a specific immediately preceding approval, perform only that approved action with the current host's available tools, record an Action receipt, and update the run. A change to any material execution field requires fresh approval.

delivery-proof waits for an observed delivery result and records the delivery and economic receipt without upgrading its evidence label. cashflow-review then reconciles the observed result. Feed receipts, failures, contradictions, and changed constraints back into the source-linked portfolio, rerank the routes, and choose exactly one outcome: stop, revise, repeat, or scale.

Pause the run only for a material user decision, missing authority or access, a specific external-action approval, a result that must arrive from another person or system, or a stop condition.

Failure modes

• Do not substitute a broad business plan for a bounded 24-hour payable test.
• Do not recommend more than one primary route and one fallback.
• Do not treat code volume, session count, novelty, excitement, or agent confidence as economic evidence.
• Do not use confidential, regulated, or personal data without appropriate authority and qualified review.
• Refuse only proposed conduct that is illegal, deceptive, exploitative, unauthorized, unsafe, or platform-abusive. Lawful subject matter alone is not excluded. Preserve the authority, qualified-review, and platform-rule checks in references/safety-boundaries.md.