@letterstory/cli 0.30.1 → 0.31.1

package/lib/cli.mjs

@@ -44,7 +44,7 @@ import {
 } from "./commands.mjs";
 
 // Keep in sync with cli/package.json.
-export const VERSION = "0.30.1";
+export const VERSION = "0.31.1";
 
 // Flags that never take a value. Listing them explicitly means `deploy get --json <id>`
 // can't accidentally swallow the id as --json's value.
@@ -571,6 +571,7 @@ Ledes (opening lines for cold email, chosen per prospect from a hand-approved li
                                                        arrive in time (fewer lines may qualify).
   lede runs [--endpoint <id>] [--limit <n>]            Recent lede calls (telemetry), newest first
   lede history [<domain>] [--days <n>] [--limit <n>]   Openers already sent to an account
+  lede evals [--sweep]                                 Lede quality evals still above zero
   lede balance                                         Ledes left on this org
 
 Landings (a link + context; auto-mentioned in articles by collections you enable):

package/lib/commands/lede.mjs

@@ -64,6 +64,10 @@ export const ARGS = {
 		endpoint_id: "endpoint",
 		limit: { flag: "limit", type: "num" },
 	},
+	// The state of the quality exercise: which evals still count above zero.
+	lede_eval_briefing: {
+		sweep: { value: (flags) => (flags.sweep === true ? true : undefined) },
+	},
 	// What has already been said to an account, so the next send does not repeat it.
 	lede_account_history: {
 		domain: { positional: 0, label: "domain" },
@@ -71,6 +75,22 @@ export const ARGS = {
 		days: { flag: "days", type: "num" },
 		limit: { flag: "limit", type: "num" },
 	},
+	// What actually happened to a lede after you sent it. The engine can measure
+	// everything up to handing the line over and nothing after, so this is the only
+	// path by which a send, a reply or a bad observation ever reaches it.
+	report_lede_outcome: {
+		run_id: { positional: 0, label: "run-id", required: true },
+		shipped_template_id: { flag: "shipped-template" },
+		sent: { value: (flags) => (flags.sent === true ? true : undefined) },
+		replied: { value: (flags) => (flags.replied === true ? true : undefined) },
+		reply_sentiment: { flag: "sentiment", enum: ["positive", "neutral", "negative"] },
+		flagged_wrong: { value: (flags) => (flags["flag-wrong"] === true ? true : undefined) },
+		flagged_reason: { flag: "reason" },
+	},
+	// Per line: picked, sent, replied, positive, flagged wrong.
+	lede_line_funnel: {
+		window_days: { flag: "days", type: "num" },
+	},
 	list_lede_endpoints: {},
 	get_lede_balance: {},
 	create_lede_endpoint: {
@@ -113,15 +133,39 @@ export async function cmdLede(ctx) {
 		case "history":
 		case "sent":
 			return ledeHistory(rest);
+		case "evals":
+		case "briefing":
+			return ledeEvals(rest);
 		case "balance":
 			return ledeBalance(rest);
+		case "outcome":
+		case "report":
+			return ledeOutcome(rest);
+		case "funnel":
+			return ledeFunnel(rest);
 		default:
 			throw new CliError(
-				`Unknown lede subcommand: ${sub ?? "(none)"}. Try: list, create, delete, endpoint, run, runs, history, balance`
+				`Unknown lede subcommand: ${sub ?? "(none)"}. Try: list, create, delete, endpoint, run, runs, history, evals, balance, outcome, funnel`
 			);
 	}
 }
 
+async function ledeOutcome(ctx) {
+	const { client, flags, positionals, io } = ctx;
+	const args = argsFromFlags(ARGS.report_lede_outcome, flags, positionals);
+	const result = await client.callTool("report_lede_outcome", args);
+	ok(ctx, `Recorded the outcome for run ${args.run_id}.`);
+	printResult(io, flags, result);
+	return 0;
+}
+
+async function ledeFunnel(ctx) {
+	const { client, flags, io } = ctx;
+	const result = await client.callTool("lede_line_funnel", argsFromFlags(ARGS.lede_line_funnel, flags));
+	printResult(io, flags, result);
+	return 0;
+}
+
 async function ledeList(ctx) {
 	const { client, flags, io } = ctx;
 	const result = await client.callTool("list_lede_endpoints", argsFromFlags(ARGS.list_lede_endpoints, flags));
@@ -241,6 +285,15 @@ async function ledeHistory(ctx) {
 	return 0;
 }
 
+// `lede evals [--sweep]` — which categories of false positive still count above zero.
+// The exercise is finished when total_hits reads zero and no new flags are arriving.
+async function ledeEvals(ctx) {
+	const { client, flags, io } = ctx;
+	const result = await client.callTool("lede_eval_briefing", argsFromFlags(ARGS.lede_eval_briefing, flags));
+	printResult(io, flags, result);
+	return 0;
+}
+
 async function ledeBalance(ctx) {
 	const { client, flags, io } = ctx;
 	const result = await client.callTool("get_lede_balance", argsFromFlags(ARGS.get_lede_balance, flags));

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@letterstory/cli",
-	"version": "0.30.1",
+	"version": "0.31.1",
 	"description": "Spin up and manage Letterstory phantom blogs from your terminal.",
 	"type": "module",
 	"license": "UNLICENSED",