Skip to main content
YAMOYAMO

Artifacts: what a run produces

The run folder and everything in it: the inputs it was given, each step of the work, the finished deliverable, and the receipt that itemises the cost.

In RunUpdated
Contents

On this page

  1. 01What the folder holds
  2. 02What you see, and what you can take away
  3. 03The receipt

Every run keeps its evidence. Not just the finished piece of work, but what it was given, what each step produced along the way, and what it cost. That collection is the run folder, and it is what makes a run explainable after the fact rather than a result you have to take on trust.

What the folder holds

  • The inputs snapshot. Exactly what this run was given, which is the first thing worth checking when a result is not what you expected.
  • Each step's document, in order. The work between the input and the deliverable. The research notes and the written report are separate documents because they were separate steps.
  • The deliverable. Identified separately from the working documents, so "the outcome" is never a matter of scrolling to the last thing.
  • The receipt.

A Playbook can also declare named outputs rather than one document, and then a run produces each of them as its own file with one marked as the primary deliverable. The Content Engine template does this: a content pack plus the article and the social posts, separately.

What you see, and what you can take away

Finished work is rendered in place. A document is rendered as a document; a data file with a table in it is rendered as a table with the first rows shown and a note about how many more there are.

Four other output kinds can be declared and are rendered as plain documents with a line saying so, because the delivery behind them is not built yet. That label is deliberate: a page that rendered an unbuilt channel as though it had been delivered would be lying about work that did not happen.

Downloads are Markdown. There is no PDF, spreadsheet or archive export.

The receipt

If the run was of your own Service, the receipt is itemised, in two groups kept deliberately apart: the model cost, at what we paid for it with no markup, one line per step; and the platform operations, priced per operation. The total is rounded up to the nearest penny, once.

If the run was of somebody else's published Service, the receipt is the price you agreed, once. There are no cost lines, because those are the platform's cost basis and the operator's margin rather than anything you bought.

Token counts are shown as a transparency readout. You are never billed per token.

Keep reading

More in Run

Keep reading

Related

See it working

Open a Playbook that does this

Every template in the library publishes its contract and a real example output.