@popoverai/dotrequirements 0.38.2 → 0.38.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
|
@@ -5,5 +5,5 @@
|
|
|
5
5
|
* request is validated against; the CLI can't import the backend, so this
|
|
6
6
|
* is that text, and a test keeps the two the same.
|
|
7
7
|
*/
|
|
8
|
-
export declare const REQUEST_CONTRACT = "Ask the person to decide something about the outcome you are drafting or reviewing. A choice presents options with a recommendation. One decision per request: if two things need deciding, ask twice.\n\nA request is JSON with a \"kind\" and the fields of that kind:\n\nchoice:\n question (text, up to 200 characters): The question, phrased as a question in the words the person will read it, not as a topic.\n options (list, at least 2): The live alternatives. More than four makes deciding harder; give more only when each is a genuine choice. Leave out anything you have already ruled out.\n name (text, up to 200 characters): The option as you would say it to the person: a noun phrase they can repeat back, not a label.\n benefits (list of text, each up to 200 characters): What is better if this option is taken. One line each, in your own words.\n costs (list of text, each up to 200 characters): What is worse if this option is taken: what it would take to build, and what the person lives with afterwards. Effort is a cost like any other.\n note (text, up to 600 characters, optional): Anything you learned or checked about this option that is not a benefit or a cost. Leave it out when there is nothing to add.\n action (object, optional): An action you may take only with the person's permission, which choosing this option gives. Once they choose it, take it with `dotreq solution approval complete <decision>`. Leave it out for anything you may do yourself.\n tool (text, up to 200 characters): The action, by name. The one that needs the person's permission is \"fix\", starting a fix.\n args (object): The action's arguments, as it takes them: a fix's are below.\n recommendation (object): Which option you would take, and the one reason why.\n option (text, up to 200 characters): The name of the option you recommend, exactly as it appears in options.\n because (text, up to 200 characters): The one reason that carries it, in one line. The full reasoning belongs in background.\n background (text, up to 4000 characters): What the person needs to know to decide well that the options do not already say. Include only context that would substantially change the quality of the decision. The person reads this after the options, so do not reiterate them.\n\nThe args of a \"fix\" action:\n name (text, up to 120 characters): A short name for the fix, from what it puts right, as \"dice below the controls\".\n words (text, up to 2000 characters): What you found, in your own words.\n plan (text, up to 2000 characters): The fix as you would put it to the person, in a sentence or two.\n redo (list, at least 1): The outcomes the fix redoes, and what would change in each.\n outcome (text, up to 200 characters): An outcome to redo, by its name.\n change (text, up to 2000 characters): What would change in it.\n rerun (list of text, each up to 200 characters): The outcomes that await the ones redone and should run again from them. Empty when none should.";
|
|
8
|
+
export declare const REQUEST_CONTRACT = "Ask the person to decide something about the outcome you are drafting or reviewing. A choice presents options with a recommendation. The person may answer in their own words instead of picking an option, and their words are then the decision you work from. One decision per request: if two things need deciding, ask twice.\n\nA request is JSON with a \"kind\" and the fields of that kind:\n\nchoice:\n question (text, up to 200 characters): The question, phrased as a question in the words the person will read it, not as a topic.\n options (list, at least 2): The live alternatives. More than four makes deciding harder; give more only when each is a genuine choice. Leave out anything you have already ruled out.\n name (text, up to 200 characters): The option as you would say it to the person: a noun phrase they can repeat back, not a label.\n benefits (list of text, each up to 200 characters): What is better if this option is taken. One line each, in your own words.\n costs (list of text, each up to 200 characters): What is worse if this option is taken: what it would take to build, and what the person lives with afterwards. Effort is a cost like any other.\n note (text, up to 600 characters, optional): Anything you learned or checked about this option that is not a benefit or a cost. Leave it out when there is nothing to add.\n action (object, optional): An action you may take only with the person's permission, which choosing this option gives. Once they choose it, take it with `dotreq solution approval complete <decision>`. Leave it out for anything you may do yourself.\n tool (text, up to 200 characters): The action, by name. The one that needs the person's permission is \"fix\", starting a fix.\n args (object): The action's arguments, as it takes them: a fix's are below.\n recommendation (object): Which option you would take, and the one reason why.\n option (text, up to 200 characters): The name of the option you recommend, exactly as it appears in options.\n because (text, up to 200 characters): The one reason that carries it, in one line. The full reasoning belongs in background.\n background (text, up to 4000 characters): What the person needs to know to decide well that the options do not already say. Include only context that would substantially change the quality of the decision. The person reads this after the options, so do not reiterate them.\n\nThe args of a \"fix\" action:\n name (text, up to 120 characters): A short name for the fix, from what it puts right, as \"dice below the controls\".\n words (text, up to 2000 characters): What you found, in your own words.\n plan (text, up to 2000 characters): The fix as you would put it to the person, in a sentence or two.\n redo (list, at least 1): The outcomes the fix redoes, and what would change in each.\n outcome (text, up to 200 characters): An outcome to redo, by its name.\n change (text, up to 2000 characters): What would change in it.\n rerun (list of text, each up to 200 characters): The outcomes that await the ones redone and should run again from them. Empty when none should.";
|
|
9
9
|
//# sourceMappingURL=contract.d.ts.map
|
|
@@ -5,7 +5,7 @@
|
|
|
5
5
|
* request is validated against; the CLI can't import the backend, so this
|
|
6
6
|
* is that text, and a test keeps the two the same.
|
|
7
7
|
*/
|
|
8
|
-
export const REQUEST_CONTRACT = `Ask the person to decide something about the outcome you are drafting or reviewing. A choice presents options with a recommendation. One decision per request: if two things need deciding, ask twice.
|
|
8
|
+
export const REQUEST_CONTRACT = `Ask the person to decide something about the outcome you are drafting or reviewing. A choice presents options with a recommendation. The person may answer in their own words instead of picking an option, and their words are then the decision you work from. One decision per request: if two things need deciding, ask twice.
|
|
9
9
|
|
|
10
10
|
A request is JSON with a "kind" and the fields of that kind:
|
|
11
11
|
|
package/package.json
CHANGED