chore: snapshot pre-v0.2.0 WIP (pathfinder/reflex/metrics/skills improvements)

Baseline for the v0.2.0 self-learning iteration. All 205 tests pass on this
state. Subsequent commits in this branch layer the knowledge base,
post-mortem coach, and persona narration on top.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

runtime/skill-metrics.js

@@ -1,27 +1,75 @@
-// Per-skill ok/fail counters. Aggregated for the lifetime of the bot
-// process (best-effort persistence is left for a future iteration —
-// today's counters reset on restart, which keeps the data store
-// simple while still being useful for incident bodies and the TUI).
+// Per-skill ok/fail counters. Persisted under state/<host>/ so the bot's
+// next run and auto-improvement prompts can learn from prior attempts,
+// not only the current process lifetime.
 
-export function createSkillMetrics() {
-	const counts = new Map(); // id → { ok, fail, lastTs }
+import fs from "node:fs";
+import path from "node:path";
+import { stateDir } from "./config.js";
 
-	function record(id, ok) {
-		const cur = counts.get(id) ?? { ok: 0, fail: 0, lastTs: 0 };
+const METRICS_PATH = path.join(stateDir, "skill-metrics.json");
+
+function loadMetrics() {
+	const counts = new Map();
+	try {
+		const raw = fs.readFileSync(METRICS_PATH, "utf8");
+		const parsed = JSON.parse(raw);
+		for (const [id, m] of Object.entries(parsed ?? {})) {
+			counts.set(id, {
+				ok: Number(m.ok ?? 0),
+				fail: Number(m.fail ?? 0),
+				lastTs: Number(m.lastTs ?? 0),
+				lastCode: m.lastCode ?? null,
+				lastDurationMs: Number(m.lastDurationMs ?? 0),
+				totalDurationMs: Number(m.totalDurationMs ?? 0),
+			});
+		}
+	} catch {}
+	return counts;
+}
+
+function saveMetrics(counts) {
+	try {
+		fs.mkdirSync(stateDir, { recursive: true });
+		const out = {};
+		for (const [id, m] of counts) out[id] = { ...m };
+		const tmp = `${METRICS_PATH}.tmp`;
+		fs.writeFileSync(tmp, JSON.stringify(out, null, 2));
+		fs.renameSync(tmp, METRICS_PATH);
+	} catch {}
+}
+
+export function createSkillMetrics({ persist = true } = {}) {
+	const counts = persist ? loadMetrics() : new Map(); // id → { ok, fail, lastTs }
+
+	function record(id, ok, { code = null, durationMs = 0 } = {}) {
+		const cur = counts.get(id) ?? { ok: 0, fail: 0, lastTs: 0, lastCode: null, lastDurationMs: 0, totalDurationMs: 0 };
 		if (ok) cur.ok++;
 		else cur.fail++;
 		cur.lastTs = Date.now();
+		cur.lastCode = code ?? cur.lastCode ?? null;
+		cur.lastDurationMs = Math.max(0, Math.round(durationMs || 0));
+		cur.totalDurationMs += cur.lastDurationMs;
 		counts.set(id, cur);
+		if (persist) saveMetrics(counts);
 	}
 
 	function snapshot() {
 		const out = {};
-		for (const [id, m] of counts) out[id] = { ...m };
+		for (const [id, m] of counts) {
+			const total = (m.ok ?? 0) + (m.fail ?? 0);
+			out[id] = {
+				...m,
+				avgDurationMs: total > 0 ? Math.round((m.totalDurationMs ?? 0) / total) : 0,
+			};
+		}
 		return out;
 	}
 
 	function reset() {
 		counts.clear();
+		if (persist) {
+			try { fs.unlinkSync(METRICS_PATH); } catch {}
+		}
 	}
 
 	return { record, snapshot, reset };