@neuroverseos/governance 0.7.0 → 0.8.1

package/dist/worldmodel-create-5SWHVNMQ.js

@@ -0,0 +1,195 @@
+import "./chunk-QWGCMQQD.js";
+
+// src/cli/worldmodel-create.ts
+import { createInterface } from "readline";
+import { writeFileSync } from "fs";
+import { resolve } from "path";
+var QUESTIONS = [
+  {
+    id: "name",
+    question: "What should we call this model? (Your name, your org, your project)",
+    placeholder: 'e.g., "Kirsten", "Auki", "My Startup"'
+  },
+  {
+    id: "mission",
+    question: "In one or two sentences \u2014 what does this system exist to do? Not a slogan. The real purpose.",
+    placeholder: 'e.g., "Protect human thinking while expanding cognitive capability through AI"'
+  },
+  {
+    id: "domains",
+    question: "What are the 2-3 big areas of focus? Not departments \u2014 the major kinds of work that matter most. Separate with commas.",
+    placeholder: 'e.g., "Safety and boundaries, Individual authority, AI as cognitive extension"'
+  },
+  {
+    id: "overlaps",
+    question: "When those areas work well TOGETHER, what does that feel like? Name a feeling for each pair.",
+    placeholder: 'e.g., "Safety + Authority = Trust, Authority + AI = Possibility"'
+  },
+  {
+    id: "center",
+    question: "When EVERYTHING is aligned \u2014 all areas working together \u2014 what does the system become? One name or phrase.",
+    placeholder: 'e.g., "The Sovereign Conduit", "Collective Vanguard Leader"'
+  },
+  {
+    id: "nonnegotiables",
+    question: "What's absolutely non-negotiable? What would you walk away over? List a few.",
+    placeholder: 'e.g., "Humans retain authority over thinking. AI extends, never replaces. People can always leave."'
+  },
+  {
+    id: "success",
+    question: `What does success look like in action? What would you point at and say "that's what I mean"?`,
+    placeholder: 'e.g., "Someone maintaining clear authorship of decisions even when AI contributed"'
+  },
+  {
+    id: "drift",
+    question: "What does drift look like? What would worry you if you saw it happening?",
+    placeholder: 'e.g., "Decision ownership quietly shifting to AI without explicit delegation"'
+  },
+  {
+    id: "priorities",
+    question: 'When tradeoffs appear, what wins? Give a few "X over Y" pairs.',
+    placeholder: 'e.g., "Safety over speed, sovereignty over convenience, diversity over uniformity"'
+  }
+];
+async function askQuestions() {
+  const rl = createInterface({
+    input: process.stdin,
+    output: process.stderr
+    // questions to stderr so stdout stays clean
+  });
+  const answers = {};
+  process.stderr.write("\n");
+  process.stderr.write("  \u2554\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2557\n");
+  process.stderr.write("  \u2551  NeuroVerseOS \u2014 Build your thinking constitution \u2551\n");
+  process.stderr.write("  \u255A\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u255D\n");
+  process.stderr.write("\n");
+  process.stderr.write("  Answer these questions in your own words.\n");
+  process.stderr.write("  There are no wrong answers \u2014 just say what you mean.\n");
+  process.stderr.write("  The AI will structure your answers into a worldmodel.\n\n");
+  for (const q of QUESTIONS) {
+    const answer = await new Promise((resolve2) => {
+      process.stderr.write(`  \x1B[1m${q.question}\x1B[0m
+`);
+      process.stderr.write(`  \x1B[2m${q.placeholder}\x1B[0m
+`);
+      rl.question("  > ", (ans) => {
+        process.stderr.write("\n");
+        resolve2(ans.trim());
+      });
+    });
+    answers[q.id] = answer;
+  }
+  rl.close();
+  return answers;
+}
+async function structureWorldmodel(answers, apiKey) {
+  const prompt = buildStructuringPrompt(answers);
+  const res = await fetch("https://api.anthropic.com/v1/messages", {
+    method: "POST",
+    headers: {
+      "x-api-key": apiKey,
+      "anthropic-version": "2023-06-01",
+      "content-type": "application/json"
+    },
+    body: JSON.stringify({
+      model: "claude-sonnet-4-20250514",
+      max_tokens: 4096,
+      system: "You are a behavioral model architect. You take conversational answers about an organization's values, purpose, and priorities, and structure them into a precise .worldmodel.md file that follows the NeuroVerseOS three-layer format. Output ONLY the markdown file content, nothing else.",
+      messages: [{ role: "user", content: prompt }]
+    })
+  });
+  if (!res.ok) {
+    throw new Error(`AI structuring failed: ${res.status}`);
+  }
+  const data = await res.json();
+  const text = data.content?.filter((c) => c.type === "text").map((c) => c.text ?? "").join("");
+  if (!text) throw new Error("AI returned no content");
+  return text.replace(/^```markdown\n?/, "").replace(/\n?```$/, "").trim();
+}
+function buildStructuringPrompt(answers) {
+  return `Structure these conversational answers into a .worldmodel.md file.
+
+The file MUST follow this exact three-layer format:
+
+---
+name: ${answers.name}
+version: 1.0.0
+---
+
+# Core Model Geometry
+
+## Mission
+(from the mission answer)
+
+## Domains
+(2-4 domains from the domains answer, each with:)
+### Domain Name
+#### Skills (8-10 skills per domain, inferred from the answers)
+#### Values (3-4 values per domain, drawn from the non-negotiables + mission)
+
+## Overlap Effects
+(from the overlaps answer, formatted as: Domain A + Domain B = Emergent State)
+
+## Center Identity
+(from the center answer)
+
+# Contextual Modifiers
+
+## Authority Layers
+(infer 4-5 authority levels appropriate to this organization)
+
+## Spatial Contexts
+(infer 4-5 contexts where behavior happens)
+
+## Interpretation Rules
+(infer 3-5 rules about how context changes meaning)
+
+# Evolution Layer
+
+## Aligned Behaviors
+(from the success answer, expanded to 5-8 items)
+
+## Drift Behaviors
+(from the drift answer, expanded to 5-8 items)
+
+## Signals
+(infer 5-7 observable signals from the answers, snake_case)
+
+## Decision Priorities
+(from the priorities answer, formatted as: preferred > secondary)
+
+## Evolution Conditions
+(infer 3-5 conditions for when the model should adapt)
+
+HERE ARE THE ANSWERS:
+
+Name: ${answers.name}
+
+Mission: ${answers.mission}
+
+Domains: ${answers.domains}
+
+Overlaps: ${answers.overlaps}
+
+Center identity: ${answers.center}
+
+Non-negotiables: ${answers.nonnegotiables}
+
+Success looks like: ${answers.success}
+
+Drift looks like: ${answers.drift}
+
+Priorities: ${answers.priorities}
+
+Output ONLY the .worldmodel.md content. No explanation. No commentary. Just the file.`;
+}
+function saveWorldmodel(content, outputPath) {
+  const resolved = resolve(outputPath);
+  writeFileSync(resolved, content, "utf-8");
+  return resolved;
+}
+export {
+  askQuestions,
+  saveWorldmodel,
+  structureWorldmodel
+};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@neuroverseos/governance",
-  "version": "0.7.0",
+  "version": "0.8.1",
   "description": "Deterministic governance engine for AI agents — enforce worlds (permanent rules) and plans (mission constraints) with full audit trace",
   "license": "Apache-2.0",
   "type": "module",