seed-example.mjs

// seed-example.mjs — build a synthetic observation store for the demo.
//
// Creates example.db with the same `observations` shape this tool reads, filled
// with FAKE data for two made-up projects. Lets anyone run the full pipeline
// (cluster → build → read) with no real memory store and no private data.
//
//   bun seed-example.mjs
//   CLAUDE_MEM_DB=./example.db MERGE_MODEL=mock bun build-digests.mjs
//   CLAUDE_MEM_DB=./example.db bun read-digests.mjs widgets-api
//
// CONFIG (env)
//   EXAMPLE_DB   output path (default: ./example.db)

import { Database } from "bun:sqlite";
import { join } from "path";

const OUT = process.env.EXAMPLE_DB || join(process.cwd(), "example.db");

const db = new Database(OUT);
db.run(`DROP TABLE IF EXISTS observations`);
db.run(`
  CREATE TABLE observations (
    id               INTEGER PRIMARY KEY,
    project          TEXT NOT NULL,
    concepts         TEXT NOT NULL,   -- JSON array of type tags
    title            TEXT NOT NULL,
    narrative        TEXT NOT NULL,
    facts            TEXT NOT NULL,   -- JSON array of strings
    created_at_epoch INTEGER NOT NULL,
    relevance_count  INTEGER DEFAULT 0
  )
`);

const TYPES = [
  "how-it-works",
  "gotcha",
  "decision",
  "pattern",
  "problem-solution",
  "what-changed",
  "trade-off",
  "why-it-exists",
];

// Two synthetic projects, deliberately lopsided so the economics are visible:
// "widgets-api" has deep history (compresses hard), "tiny-cli" is shallow
// (near the floor where consolidation barely helps).
const PLAN = {
  "widgets-api": 220,
  "tiny-cli": 12,
};

const insert = db.query(`
  INSERT INTO observations
    (project, concepts, title, narrative, facts, created_at_epoch, relevance_count)
  VALUES (?, ?, ?, ?, ?, ?, 0)
`);

let epoch = Date.UTC(2026, 0, 1); // Jan 1 2026, marches forward per row
let total = 0;

for (const [project, n] of Object.entries(PLAN)) {
  for (let i = 0; i < n; i++) {
    const type = TYPES[i % TYPES.length];
    const title = `${project}: ${type} note #${i + 1}`;
    const narrative =
      `During session ${i + 1} on ${project}, work touching the ${type} ` +
      `area produced a distilled memory. Several of these overlap on purpose, ` +
      `so consolidation has duplicates to collapse and stale points to drop.`;
    const facts = JSON.stringify([
      `${type} fact A for ${project} #${i + 1}`,
      `${type} fact B for ${project} #${i + 1}`,
    ]);
    const concepts = JSON.stringify([type]);
    epoch += 60_000; // +1 minute per observation, strictly increasing
    insert.run(project, concepts, title, narrative, facts, epoch);
    total++;
  }
}

db.close();
console.log(`Seeded ${total} synthetic observations → ${OUT}`);
console.log(`Projects: ${Object.entries(PLAN).map(([p, n]) => `${p}(${n})`).join(", ")}`);
console.log(`\nNext:`);
console.log(`  CLAUDE_MEM_DB=./example.db MERGE_MODEL=mock bun build-digests.mjs`);
console.log(`  CLAUDE_MEM_DB=./example.db bun read-digests.mjs widgets-api`);