ms-ai-architect/scripts/kb-update/lib/detection-schedule.mjs

// detection-schedule.mjs — Spor C / C1: opt-in session-anchored detection.
//
// The DETECTION layer (poll → report → discover → detect-skill-lifecycle) is
// pure node: it polls Microsoft Learn sitemaps and writes JSON reports, and
// NEVER contacts Anthropic/Claude. It is therefore OUTSIDE the ToS surface —
// Consumer Terms §3.7 restricts automated access to "our Services" (Anthropic),
// not running local scripts. The APPLY layer (which invokes Claude) stays
// manual, in-session, and gated. This module is the pure decision core for the
// SessionStart hook + the run-detection.mjs executor.
//
// Zero dependencies beyond node:fs/path. parseScheduleConfig is pure.

import { readFileSync, existsSync } from 'node:fs';
import { join } from 'node:path';

// Opt-in: detection is OFF until the user sets enabled:true. "Frivillig å sette
// opp — men gjøres det, skal det virke." (operator 2026-06-20)
export const DEFAULT_SCHEDULE_CONFIG = Object.freeze({
  enabled: false,
  interval_days: 7,
  include_skill_lifecycle: true,
  // Tier 2 (launchd) cadence. 'daily' = poll on every (daily) fire; 'interval'
  // = throttle to interval_days via the same staleness gate Tier 1 uses. Read
  // ONLY by the Tier-2 entrypoint (scheduler.mjs run); the Tier-1 hook ignores
  // it, so existing behavior is unchanged. Operator default: daily.
  os_scheduler_cadence: 'daily',
});

// The ordered, Claude-FREE detection pipeline. Every step is a local node
// script that produces a JSON report; none invoke Claude. run-detection.mjs
// resolves each as `node <pluginRoot>/scripts/<dir>/<script> <args>`.
export const DETECTION_STEPS = Object.freeze([
  { name: 'poll-sitemaps', dir: 'kb-update', script: 'poll-sitemaps.mjs', args: ['--force'] },
  { name: 'report-changes', dir: 'kb-update', script: 'report-changes.mjs', args: [] },
  { name: 'discover-new-urls', dir: 'kb-update', script: 'discover-new-urls.mjs', args: ['--limit', '500'] },
  { name: 'detect-skill-lifecycle', dir: 'kb-eval', script: 'detect-skill-lifecycle.mjs', args: ['--write'], skillLifecycle: true },
]);

const CONFIG_FILENAME = 'ms-ai-architect.local.md';

/** Coerce a YAML-ish scalar to bool/number/string. Unknown booleans => false. */
function coerce(key, raw) {
  const v = raw.trim();
  if (key === 'enabled' || key === 'include_skill_lifecycle') {
    if (v === 'true') return true;
    if (v === 'false') return false;
    return undefined; // non-boolean => caller keeps default (or OFF for enabled)
  }
  if (key === 'interval_days') {
    const n = Number.parseInt(v, 10);
    return Number.isFinite(n) && n > 0 ? n : undefined;
  }
  if (key === 'os_scheduler_cadence') {
    // Only 'interval' opts out of daily; any other/garbage value => default daily.
    return v === 'interval' ? 'interval' : 'daily';
  }
  return undefined;
}

/**
 * Parse the `scheduled_detection:` block out of a config file's text. Pure;
 * tolerant: any parse failure or missing block yields the OFF default. Reads a
 * minimal indented key: value sub-block (no YAML dependency — the format is
 * documented + controlled).
 * @param {string} text
 * @returns {{enabled: boolean, interval_days: number, include_skill_lifecycle: boolean}}
 */
export function parseScheduleConfig(text) {
  const cfg = { ...DEFAULT_SCHEDULE_CONFIG };
  if (typeof text !== 'string' || !text.includes('scheduled_detection:')) return cfg;
  const lines = text.split('\n');
  let inBlock = false;
  for (const line of lines) {
    if (/^\s*scheduled_detection:\s*$/.test(line)) { inBlock = true; continue; }
    if (!inBlock) continue;
    // Block ends at the first non-indented, non-blank line (e.g. `---` or next key).
    if (line.trim() === '') continue;
    if (!/^\s+\S/.test(line)) break;
    const m = line.match(/^\s+([A-Za-z_]+):\s*(.*)$/);
    if (!m) continue;
    const [, key, raw] = m;
    if (!(key in DEFAULT_SCHEDULE_CONFIG)) continue;
    const val = coerce(key, raw);
    if (val !== undefined) cfg[key] = val;
  }
  return cfg;
}

/**
 * Load the opt-in schedule config from <pluginRoot>/ms-ai-architect.local.md
 * (gitignored per the plugin-settings pattern). Absent/unreadable => OFF default.
 * @param {string} pluginRoot
 * @returns {{enabled: boolean, interval_days: number, include_skill_lifecycle: boolean}}
 */
export function loadScheduleConfig(pluginRoot) {
  const path = join(pluginRoot, CONFIG_FILENAME);
  if (!existsSync(path)) return { ...DEFAULT_SCHEDULE_CONFIG };
  try {
    return parseScheduleConfig(readFileSync(path, 'utf8'));
  } catch {
    return { ...DEFAULT_SCHEDULE_CONFIG };
  }
}

/**
 * Decide whether the SessionStart hook should background-spawn detection. Pure.
 * Opt-in: a disabled config never runs (the default). Enabled + stale (days
 * since last poll >= interval) runs; enabled + fresh does not. force overrides.
 * @param {{enabled: boolean, interval_days: number}} config
 * @param {number} lastPollDaysAgo  Infinity when never polled
 * @param {{force?: boolean}} [opts]
 * @returns {{run: boolean, reason: string}}
 */
export function shouldRunDetection(config, lastPollDaysAgo, opts = {}) {
  if (opts.force) return { run: true, reason: 'force' };
  if (!config || !config.enabled) return { run: false, reason: 'disabled (opt-in off)' };
  const interval = config.interval_days ?? DEFAULT_SCHEDULE_CONFIG.interval_days;
  if (lastPollDaysAgo >= interval) {
    return { run: true, reason: `stale (${Math.floor(lastPollDaysAgo)}d >= ${interval}d)` };
  }
  return { run: false, reason: 'fresh (within interval)' };
}

/**
 * Days since the last sitemap poll, read from a parsed change-report.json.
 * Pure; mirrors the SessionStart hook's computation. Missing/invalid
 * last_poll => Infinity ("infinitely stale" => the 'interval' cadence runs).
 * @param {object|null} report  parsed change-report.json ({ last_poll?: string })
 * @param {number} now  Date.now()
 * @returns {number}
 */
export function daysSinceLastPoll(report, now) {
  const DAY_MS = 24 * 60 * 60 * 1000;
  if (!report || !report.last_poll) return Infinity;
  const t = new Date(report.last_poll).getTime();
  if (!Number.isFinite(t)) return Infinity;
  return (now - t) / DAY_MS;
}

/**
 * Build a compact one-liner summarizing skill-lifecycle signals for the
 * SessionStart hook. Pure; tolerant of partial/missing reports. Returns null
 * when there is nothing worth surfacing.
 * @param {object|null} report  parsed skill-lifecycle-report.json
 * @param {{threshold?: number}} [opts]  K10 overlap threshold (default 7.0)
 * @returns {string|null}
 */
export function summarizeSkillLifecycle(report, opts = {}) {
  if (!report || typeof report !== 'object') return null;
  const threshold = opts.threshold ?? 7.0;
  const overlapFail = (report.overlap?.pairs ?? []).filter((p) => (p?.combined ?? 0) >= threshold).length;
  const gaps = report.coverage?.summary?.gaps ?? (report.coverage?.gaps ?? []).length;
  const thin = report.coverage?.summary?.thin ?? 0;
  const bloat = (report.bloat?.summary?.bloatCandidates ?? []).length;
  const stale = (report.bloat?.summary?.staleCandidates ?? []).length;

  const parts = [];
  if (overlapFail > 0) parts.push(`${overlapFail} overlapp-par`);
  if (gaps > 0) parts.push(`${gaps} gap`);
  if (thin > 0) parts.push(`${thin} tynne`);
  if (bloat > 0) parts.push(`${bloat} bloat`);
  if (stale > 0) parts.push(`${stale} stale`);
  return parts.length ? `Skill-signaler: ${parts.join(', ')}` : null;
}