`, favicon, and body of `Root()` to\nmatch your app's existing `index.html`, and stamp anything else your shell\nreferences directly (images, css) with `immutableFileUrl()`.\n\n1. Create `frontend/root.tsx`:\n\n```tsx\n/** @jsxImportSource npm:hono@4/jsx */\nimport { raw } from \"npm:hono@4/html\";\nimport { immutableFileUrl } from \"https://esm.town/v/std/utils/index.ts\";\n\n// The HTML shell for the app. It is never cached, so immutableFileUrl can stamp\n// the current val version into each asset URL — the browser caches those\n// immutably, and publishing the val changes the URLs, refetching exactly once.\nexport function Root() {\n return (\n <>\n {raw(\"<!DOCTYPE html>\")}\n <html lang=\"en\">\n <head>\n <meta charSet=\"UTF-8\" />\n <meta name=\"viewport\" content=\"width=device-width, initial-scale=1.0\" />\n <title>React Hono Val Town Starter

[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_388":389,"_390":391},"routes/_app.x.$handle.$name",{"_11":12,"_13":14,"_15":16,"_18":19,"_17":-7,"_293":-5,"_20":294,"_22":295,"_296":297,"_298":299,"_26":300,"_24":301,"_302":-5,"_303":-5,"_304":70,"_305":306,"_307":64,"_308":-5,"_309":-5,"_310":311},"routes/_app.x.$handle.$name.branch.$branch.code.$",{"_11":12,"_13":14,"_15":16,"_17":-7,"_18":19,"_20":21,"_22":23,"_24":25,"_26":27,"_28":-5,"_29":-5,"_30":-5,"_31":-5,"_32":-5},"handle","jonasolof","name","training-coach","branchName","main","version","path","intervals_sync.ts","project",{"_33":235,"_13":14,"_45":236,"_237":-5,"_238":70,"_239":240,"_241":240,"_242":243,"_244":-5,"_245":246,"_247":79,"_248":64,"_249":-5,"_250":-5,"_251":127,"_252":253,"_254":255,"_256":257,"_258":259,"_260":-5,"_261":262,"_263":262,"_264":265,"_266":265,"_267":265,"_268":262,"_269":262,"_270":271,"_272":273,"_274":275,"_276":70},"tree",[51,52,53,54,55,56,57,58],"file",{"_33":34,"_35":36,"_37":38,"_13":19,"_39":40,"_41":42,"_17":43,"_44":-5,"_45":46,"_47":48,"_49":-5,"_18":50},"contents",[],"upstream","pull","ahead","behind","envVars","id","49d4ae77-f2b4-4ed9-88eb-15857eeeaf66","branchId","60471d74-2b50-43e0-a779-1eff6bc1aaa3","fileId","30c78390-9f2a-11f1-af76-1607ee4eb77e","content","import { sqlite } from \"https://esm.town/v/std/sqlite/main.ts\";\n\ntype IntervalsActivity = {\n id?: string;\n start_date_local?: string;\n start_date?: string;\n name?: string;\n type?: string;\n moving_time?: number;\n elapsed_time?: number;\n distance?: number;\n icu_distance?: number;\n average_heartrate?: number;\n max_heartrate?: number;\n file_type?: string;\n};\n\ntype NormalizedActivity = {\n id: string;\n localDate: string;\n startedAt: string;\n completedAt: string;\n name: string;\n type: string;\n durationS: number;\n distanceKm: number | null;\n avgHr: number | null;\n maxHr: number | null;\n};\n\nconst TZ = \"Europe/Stockholm\";\n\nfunction isoDate(date: Date) {\n return date.toISOString().slice(0, 10);\n}\n\nfunction timezoneOffsetMinutes(date: Date, timeZone: string) {\n const parts = new Intl.DateTimeFormat(\"en-CA\", {\n timeZone,\n year: \"numeric\",\n month: \"2-digit\",\n day: \"2-digit\",\n hour: \"2-digit\",\n minute: \"2-digit\",\n second: \"2-digit\",\n hourCycle: \"h23\",\n }).formatToParts(date);\n const value = (type: string) => Number(parts.find((p) => p.type === type)?.value ?? 0);\n const asUtc = Date.UTC(value(\"year\"), value(\"month\") - 1, value(\"day\"), value(\"hour\"), value(\"minute\"), value(\"second\"));\n return Math.round((asUtc - date.getTime()) / 60000);\n}\n\nfunction stockholmLocalToUtc(local: string) {\n const match = local.match(/^(\\d{4})-(\\d{2})-(\\d{2})T(\\d{2}):(\\d{2}):(\\d{2})/);\n if (!match) return new Date(local).toISOString();\n const [, y, m, d, hh, mm, ss] = match;\n const naiveUtcMs = Date.UTC(Number(y), Number(m) - 1, Number(d), Number(hh), Number(mm), Number(ss));\n let guess = new Date(naiveUtcMs);\n let offset = timezoneOffsetMinutes(guess, TZ);\n guess = new Date(naiveUtcMs - offset * 60_000);\n offset = timezoneOffsetMinutes(guess, TZ);\n return new Date(naiveUtcMs - offset * 60_000).toISOString();\n}\n\nfunction normalize(a: IntervalsActivity): NormalizedActivity | null {\n if (!a.id || !a.start_date_local) return null;\n const durationS = Math.round(a.moving_time ?? a.elapsed_time ?? 0);\n if (durationS <= 0) return null;\n const startedAt = a.start_date ? new Date(a.start_date).toISOString() : stockholmLocalToUtc(a.start_date_local);\n const completedAt = new Date(new Date(startedAt).getTime() + durationS * 1000).toISOString();\n const distanceMeters = typeof a.icu_distance === \"number\" ? a.icu_distance : a.distance;\n return {\n id: a.id,\n localDate: a.start_date_local.slice(0, 10),\n startedAt,\n completedAt,\n name: a.name || a.type || \"Garmin activity\",\n type: a.type || \"Activity\",\n durationS,\n distanceKm: typeof distanceMeters === \"number\" ? Math.round(distanceMeters / 10) / 100 : null,\n avgHr: typeof a.average_heartrate === \"number\" ? Math.round(a.average_heartrate) : null,\n maxHr: typeof a.max_heartrate === \"number\" ? Math.round(a.max_heartrate) : null,\n };\n}\n\nasync function fetchRecentIntervals(days = 21) {\n const apiKey = Deno.env.get(\"INTERVALS_API_KEY\");\n if (!apiKey) throw new Error(\"Missing INTERVALS_API_KEY environment variable\");\n\n const newest = new Date();\n const oldest = new Date(newest);\n oldest.setUTCDate(oldest.getUTCDate() - days);\n const url = new URL(\"https://intervals.icu/api/v1/athlete/0/activities\");\n url.searchParams.set(\"oldest\", isoDate(oldest));\n url.searchParams.set(\"newest\", isoDate(newest));\n\n const response = await fetch(url, {\n headers: { Authorization: `Basic ${btoa(`API_KEY:${apiKey}`)}` },\n });\n if (!response.ok) {\n const body = await response.text();\n throw new Error(`Intervals.icu ${response.status}: ${body.slice(0, 500)}`);\n }\n return ((await response.json()) as IntervalsActivity[]).map(normalize).filter(Boolean) as NormalizedActivity[];\n}\n\nasync function ensureSchema() {\n await sqlite.execute(`CREATE TABLE IF NOT EXISTS external_activity_links (\n provider TEXT NOT NULL,\n external_id TEXT NOT NULL,\n session_id TEXT NOT NULL,\n linked_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n PRIMARY KEY(provider, external_id),\n FOREIGN KEY(session_id) REFERENCES sessions(id) ON DELETE CASCADE\n )`);\n}\n\nasync function linkedSessionId(externalId: string) {\n const result = await sqlite.execute({\n sql: \"SELECT session_id FROM external_activity_links WHERE provider = ? AND external_id = ? LIMIT 1\",\n args: [\"intervals_icu\", externalId],\n });\n return result.rows[0]?.session_id as string | undefined;\n}\n\nasync function findExistingSession(activity: NormalizedActivity) {\n const startMs = new Date(activity.startedAt).getTime();\n const windowStart = new Date(startMs - 24 * 60 * 60 * 1000).toISOString();\n const windowEnd = new Date(startMs + 24 * 60 * 60 * 1000).toISOString();\n const result = await sqlite.execute({\n sql: `SELECT id, name, started_at, duration_seconds, distance_km, avg_hr, source\n FROM sessions\n WHERE started_at BETWEEN ? AND ?\n ORDER BY started_at ASC`,\n args: [windowStart, windowEnd],\n });\n\n const candidates = result.rows as Array>;\n let best: { id: string; score: number } | null = null;\n for (const row of candidates) {\n const duration = Number(row.duration_seconds ?? 0);\n const distance = row.distance_km == null ? null : Number(row.distance_km);\n const rowStartMs = new Date(String(row.started_at)).getTime();\n const timeDiffMinutes = Math.abs(rowStartMs - startMs) / 60_000;\n const source = String(row.source ?? \"\");\n let score = Math.abs(duration - activity.durationS) / 30;\n\n if (source === \"val_town\") {\n if (timeDiffMinutes > 120) continue;\n score += timeDiffMinutes / 5;\n }\n\n if (activity.distanceKm != null) {\n if (distance == null) continue;\n const distanceDiff = Math.abs(distance - activity.distanceKm);\n if (distanceDiff > 0.25) continue;\n score += distanceDiff * 20;\n } else if (Math.abs(duration - activity.durationS) > 240) {\n continue;\n }\n\n if (Math.abs(duration - activity.durationS) > 300) continue;\n if (!best || score < best.score) best = { id: String(row.id), score };\n }\n return best?.id;\n}\n\nasync function enrichAndLink(sessionId: string, activity: NormalizedActivity) {\n await sqlite.batch([\n {\n sql: `UPDATE sessions SET\n duration_seconds = CASE WHEN duration_seconds = 0 THEN ? ELSE duration_seconds END,\n distance_km = COALESCE(distance_km, ?),\n avg_hr = COALESCE(avg_hr, ?)\n WHERE id = ?`,\n args: [activity.durationS, activity.distanceKm, activity.avgHr, sessionId],\n },\n {\n sql: `INSERT OR IGNORE INTO external_activity_links (provider, external_id, session_id)\n VALUES (?, ?, ?)`,\n args: [\"intervals_icu\", activity.id, sessionId],\n },\n ...(activity.maxHr == null ? [] : [{\n sql: `INSERT OR REPLACE INTO performance_entries (id, session_id, metric_key, value, unit)\n VALUES (?, ?, 'max_hr', ?, 'bpm')`,\n args: [`intervals:${activity.id}:max_hr`, sessionId, activity.maxHr],\n }]),\n ]);\n}\n\nasync function createSession(activity: NormalizedActivity) {\n const sessionId = `intervals-${activity.id}`;\n const isRun = activity.type.toLowerCase().includes(\"run\");\n const pace = isRun && activity.distanceKm && activity.distanceKm > 0\n ? Math.round(activity.durationS / activity.distanceKm)\n : null;\n\n await sqlite.batch([\n {\n sql: `INSERT OR IGNORE INTO sessions (\n id, client_id, template_id, name, type, started_at, completed_at,\n duration_seconds, notes, distance_km, avg_pace_sec_km, avg_hr,\n source, source_external_id, sync_status, processed_at\n ) VALUES (?, ?, NULL, ?, 'open', ?, ?, ?, ?, ?, ?, ?, 'intervals_icu', ?, 'official', NULL)`,\n args: [\n sessionId,\n sessionId,\n activity.name,\n activity.startedAt,\n activity.completedAt,\n activity.durationS,\n `Auto-importerad från Intervals.icu (${activity.type}).`,\n activity.distanceKm,\n pace,\n activity.avgHr,\n `intervals:${activity.id}`,\n ],\n },\n {\n sql: `INSERT OR IGNORE INTO external_activity_links (provider, external_id, session_id)\n VALUES (?, ?, ?)`,\n args: [\"intervals_icu\", activity.id, sessionId],\n },\n ...(activity.maxHr == null ? [] : [{\n sql: `INSERT OR REPLACE INTO performance_entries (id, session_id, metric_key, value, unit)\n VALUES (?, ?, 'max_hr', ?, 'bpm')`,\n args: [`intervals:${activity.id}:max_hr`, sessionId, activity.maxHr],\n }]),\n ]);\n return sessionId;\n}\n\nexport async function syncIntervals() {\n await ensureSchema();\n const activities = await fetchRecentIntervals(21);\n const summary = { fetched: activities.length, created: 0, matched: 0, alreadyLinked: 0, sessions: [] as Array> };\n\n for (const activity of activities) {\n const linked = await linkedSessionId(activity.id);\n if (linked) {\n await enrichAndLink(linked, activity);\n summary.alreadyLinked += 1;\n continue;\n }\n\n const existing = await findExistingSession(activity);\n if (existing) {\n await enrichAndLink(existing, activity);\n summary.matched += 1;\n summary.sessions.push({ activity: activity.id, action: \"matched\", sessionId: existing, name: activity.name });\n continue;\n }\n\n const sessionId = await createSession(activity);\n summary.created += 1;\n summary.sessions.push({ activity: activity.id, action: \"created\", sessionId, name: activity.name });\n }\n\n await sqlite.execute({\n sql: `INSERT INTO sync_state (key, value, updated_at) VALUES ('intervals_last_sync', ?, CURRENT_TIMESTAMP)\n ON CONFLICT(key) DO UPDATE SET value = excluded.value, updated_at = CURRENT_TIMESTAMP`,\n args: [JSON.stringify({ at: new Date().toISOString(), ...summary })],\n });\n\n console.log(JSON.stringify(summary, null, 2));\n return summary;\n}\n\nexport default async function (_interval?: Interval) {\n return await syncIntervals();\n}\n","fileType","interval",10,"parentId","createdAt",["D",1787515262921],"branchVersion",55,"label",[19],{"_33":123,"_13":124,"_37":125,"_44":-5,"_35":36,"_47":66,"_63":64,"_45":126,"_17":127,"_41":128,"_39":-5,"_69":70,"_49":-5,"_71":129,"_18":130},{"_33":114,"_13":115,"_37":116,"_44":-5,"_35":36,"_47":117,"_63":64,"_45":118,"_17":119,"_41":24,"_39":120,"_69":70,"_49":-5,"_71":121,"_18":122},{"_33":106,"_13":107,"_37":108,"_44":-5,"_35":36,"_47":109,"_63":64,"_45":110,"_17":66,"_41":24,"_39":111,"_69":70,"_49":-5,"_71":112,"_18":113},{"_33":97,"_13":98,"_37":99,"_44":-5,"_35":36,"_47":100,"_63":64,"_45":101,"_17":102,"_41":24,"_39":103,"_69":70,"_49":-5,"_71":104,"_18":105},{"_33":88,"_13":89,"_37":90,"_44":-5,"_35":36,"_47":91,"_63":64,"_45":92,"_17":93,"_41":67,"_39":94,"_69":70,"_49":-5,"_71":95,"_18":96},{"_33":77,"_13":78,"_37":79,"_44":-5,"_35":36,"_47":80,"_63":64,"_45":81,"_17":82,"_41":83,"_39":84,"_69":70,"_49":85,"_71":86,"_18":87},{"_33":34,"_13":19,"_37":38,"_44":-5,"_35":36,"_47":48,"_63":64,"_45":74,"_17":43,"_41":42,"_39":40,"_69":70,"_49":-5,"_71":75,"_18":76},{"_33":59,"_13":60,"_37":61,"_44":-5,"_35":36,"_47":62,"_63":64,"_45":65,"_17":66,"_41":67,"_39":68,"_69":70,"_49":-5,"_71":72,"_18":73},"d30e846d-4677-44f0-9d7a-a14acb06a14e","smoke.ts","7f6e32a4-9e72-11f1-9f94-1607ee4eb77e",19,"latest",true,["D",1787435016118],1,"script","import { getBootstrapData } from \"./db.ts\";\n\nconst data = await getBootstrapData();\nconst result = {\n templates: data.templates.length,\n sessions: data.sessions.length,\n recommendationModel: data.recommendation.model,\n pendingSync: data.progress.pending_sync,\n hasFrequency: Array.isArray(data.progress.frequency_by_week),\n};\nif (result.templates < 8 || result.sessions < 1) throw new Error(\"Seed data missing\");\nconsole.log(result);\nexport default result;","deleted",false,"children",[],[60],["D",1787515262921],[],[19],"993edeb7-9108-4a6a-a581-3b19386bf19f","index.ts","01a02b61-7568-77b9-9da3-bedb8e6ed0b5",16,["D",1787434890404],3,"http","import { Hono } from \"npm:hono@4\";\nimport { getOAuthUserData, oauthMiddleware } from \"https://esm.town/v/std/oauth/middleware.ts\";\nimport { parseVal, serveImmutableFile } from \"https://esm.town/v/std/utils/index.ts\";\nimport { acknowledgeCoachSessions, createSession, ensureSchema, getBootstrapData, getCoachContext } from \"./db.ts\";\nimport { Root } from \"./frontend/root.tsx\";\n\nconst app = new Hono();\nconst OWNER = \"jonasolof\";\n\nasync function viewer(req: Request) {\n const session = await getOAuthUserData(req);\n return session?.user?.username === OWNER ? session.user : null;\n}\n\napp.get(\"/health\", (c) => c.json({ ok: true, app: \"training-coach\", database: \"val-scoped-sqlite\" }));\n\napp.get(\"/\", async (c) => {\n const session = await getOAuthUserData(c.req.raw);\n if (!session?.user) return c.redirect(\"/auth/login\");\n if (session.user.username !== OWNER) return c.text(\"Den här träningsappen är personlig.\", 403);\n await ensureSchema();\n return c.html(Root());\n});\n\napp.get(\"/manifest.webmanifest\", (c) => c.json({\n name: \"Jonas träning\",\n short_name: \"Träning\",\n start_url: \"/\",\n display: \"standalone\",\n background_color: \"#f3f1eb\",\n theme_color: \"#17211b\",\n icons: [{ src: \"/frontend/favicon.svg\", sizes: \"any\", type: \"image/svg+xml\" }],\n}));\n\napp.use(\"/api/*\", async (c, next) => {\n if (!(await viewer(c.req.raw))) return c.json({ error: \"Logga in med rätt Val Town-konto\" }, 401);\n await next();\n});\n\napp.get(\"/api/bootstrap\", async (c) => c.json(await getBootstrapData()));\n\napp.post(\"/api/sessions\", async (c) => {\n try {\n const session = await createSession(await c.req.json());\n return c.json(session, 201);\n } catch (error) {\n return c.json({ error: error instanceof Error ? error.message : \"Passet kunde inte sparas\" }, 400);\n }\n});\n\napp.get(\"/api/coach/context\", async (c) => c.json(await getCoachContext(c.req.query(\"since\"))));\n\napp.post(\"/api/coach/ack\", async (c) => {\n const body = await c.req.json<{ ids?: string[] }>();\n return c.json(await acknowledgeCoachSessions(Array.isArray(body.ids) ? body.ids : []));\n});\n\napp.get(\"/__immutable/*\", (c) => serveImmutableFile(c.req.path));\napp.get(\"/frontend/*\", (c) => serveImmutableFile(c.req.path));\napp.get(\"/source\", (c) => c.redirect(parseVal().links.self.val));\n\napp.notFound((c) => c.redirect(\"/\"));\napp.onError((err) => Promise.reject(err));\n\nexport default oauthMiddleware(app.fetch);\n","jonas-training",[],[78],"7ef95792-9656-447b-9daa-2fff1d85865f","db.ts","fde5ebb4-9e71-11f1-9141-1607ee4eb77e",39,["D",1787437870810],4,"import { sqlite } from \"https://esm.town/v/std/sqlite/main.ts\";\n\nexport type SessionInput = {\n client_id?: string;\n template_id?: string | null;\n name: string;\n type: \"amrap\" | \"emom\" | \"for_time\" | \"open\";\n started_at: string;\n completed_at?: string;\n duration_seconds: number;\n rounds?: number | null;\n partial_reps?: number | null;\n effort?: number | null;\n feeling?: number | null;\n notes?: string;\n weight_kg?: number | null;\n distance_km?: number | null;\n avg_pace_sec_km?: number | null;\n avg_hr?: number | null;\n cadence_spm?: number | null;\n laps?: Array<{ elapsed_seconds: number; reps?: number | null; note?: string }>;\n performance?: Array<{ metric_key: string; value: number; unit?: string }>;\n};\n\nconst templates = [\n {\n id: \"kb-standard-20\",\n name: \"Standardpass KB 20\",\n type: \"amrap\",\n duration: 1200,\n rounds: null,\n weight: 20,\n description: \"Ett enkelt helkroppspass med puls – samma kettlebell hela vägen.\",\n tags: [\"Kettlebell\", \"Puls\", \"Standard\"],\n isStandard: 1,\n exercises: [\n { name: \"Pop squats\", reps: \"15\" },\n { name: \"Armhävningar\", reps: \"10\" },\n { name: \"Reverse lunges\", reps: \"8 / sida\" },\n { name: \"Swings\", reps: \"15\" },\n { name: \"Floor press\", reps: \"5 / sida\" },\n ],\n },\n {\n id: \"emom-pulse-press\",\n name: \"EMOM Puls + press\",\n type: \"emom\",\n duration: 1200,\n rounds: 4,\n weight: 20,\n description: \"Fem tydliga minuter som upprepas fyra gånger. Press är den avsiktliga svaga länken.\",\n tags: [\"EMOM\", \"Kettlebell\", \"Press\"],\n isStandard: 1,\n exercises: [\n { name: \"Swings\", reps: \"15\" },\n { name: \"Armhävningar\", reps: \"12\" },\n { name: \"Pop squats\", reps: \"12\" },\n { name: \"Reverse lunges\", reps: \"10 / sida\" },\n { name: \"Floor press\", reps: \"5 / sida\" },\n ],\n },\n {\n id: \"core-stability\",\n name: \"Core & stabilitet\",\n type: \"amrap\",\n duration: 900,\n rounds: null,\n weight: 16,\n description: \"Bål, axelstabilitet och kontroll på 15 minuter.\",\n tags: [\"Core\", \"Stabilitet\"],\n isStandard: 0,\n exercises: [\n { name: \"Forearm plank\", reps: \"Till failure\" },\n { name: \"Around the worlds\", reps: \"10 / håll\" },\n { name: \"Horizontal swings\", reps: \"10\" },\n { name: \"Overhead marches\", reps: \"10 / sida\" },\n { name: \"Kneeling wood chops\", reps: \"6 / sida\" },\n { name: \"Plank pull-through + push-up\", reps: \"10\" },\n ],\n },\n {\n id: \"kb-complex\",\n name: \"Kettlebell-komplex\",\n type: \"amrap\",\n duration: 1200,\n rounds: null,\n weight: 20,\n description: \"Det ursprungliga Pass 2 från träningsloggen – kort, funktionellt och pulshöjande.\",\n tags: [\"Kettlebell\", \"Puls\"],\n isStandard: 0,\n exercises: [\n { name: \"Halos + goblet squats\", reps: \"6 + 2\" },\n { name: \"Alternating single-arm swings\", reps: \"20\" },\n { name: \"Reverse lunges + forearm curl\", reps: \"10\" },\n { name: \"Burpees over the bell\", reps: \"10\" },\n ],\n },\n {\n id: \"functional-pulse\",\n name: \"Funktionell styrka & puls\",\n type: \"for_time\",\n duration: 0,\n rounds: 6,\n weight: 16,\n description: \"Sex varv på tid med 1–2 minuters vila mellan varven.\",\n tags: [\"Kettlebell\", \"Puls\", \"På tid\"],\n isStandard: 0,\n exercises: [\n { name: \"KB jump hops\", reps: \"20\" },\n { name: \"Alternating offset push-ups\", reps: \"12\" },\n { name: \"Squat + pullovers\", reps: \"12\" },\n { name: \"Single-arm front rack lunges\", reps: \"6 / sida\" },\n { name: \"Rotational KB uppercuts\", reps: \"6 / sida\" },\n ],\n },\n {\n id: \"clean-press-squat\",\n name: \"Clean, press & squat\",\n type: \"amrap\",\n duration: 900,\n rounds: null,\n weight: 16,\n description: \"Teknik och presskapacitet. Byt clean om handleden protesterar.\",\n tags: [\"Kettlebell\", \"Press\", \"Teknik\"],\n isStandard: 0,\n exercises: [\n { name: \"Cleans\", reps: \"5\" },\n { name: \"Presses\", reps: \"5\" },\n { name: \"Goblet squats\", reps: \"5\" },\n ],\n },\n {\n id: \"run-zone-2\",\n name: \"Lugn löpning\",\n type: \"open\",\n duration: 1800,\n rounds: null,\n weight: null,\n description: \"30 minuter i zon 2. Jämn känsla före fart.\",\n tags: [\"Löpning\", \"Zon 2\"],\n isStandard: 1,\n exercises: [{ name: \"Löpning zon 2\", reps: \"30 min\" }],\n },\n {\n id: \"ttb-prep\",\n name: \"Toes-to-bar prep\",\n type: \"amrap\",\n duration: 900,\n rounds: null,\n weight: null,\n description: \"Axlar och bål inför längdskidor och framtida toes-to-bar progression.\",\n tags: [\"Core\", \"Skills\", \"Toes-to-bar\"],\n isStandard: 0,\n exercises: [\n { name: \"Toes-to-bar / hanging leg raises\", reps: \"6–8\" },\n { name: \"Dead hang\", reps: \"30 sek\" },\n { name: \"Sit-ups\", reps: \"12\" },\n { name: \"Plank pull-through\", reps: \"8 / sida\" },\n ],\n },\n];\n\nlet schemaReady: Promise | null = null;\n\nexport function ensureSchema() {\n if (!schemaReady) schemaReady = setupSchema();\n return schemaReady;\n}\n\nasync function setupSchema() {\n await sqlite.batch([\n { sql: `CREATE TABLE IF NOT EXISTS workout_templates (\n id TEXT PRIMARY KEY, name TEXT NOT NULL, type TEXT NOT NULL,\n duration_seconds INTEGER NOT NULL DEFAULT 0, rounds_target INTEGER,\n default_weight_kg REAL, description TEXT NOT NULL DEFAULT '',\n tags_json TEXT NOT NULL DEFAULT '[]', exercises_json TEXT NOT NULL DEFAULT '[]',\n is_standard INTEGER NOT NULL DEFAULT 0, active INTEGER NOT NULL DEFAULT 1,\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )` },\n { sql: `CREATE TABLE IF NOT EXISTS sessions (\n id TEXT PRIMARY KEY, client_id TEXT UNIQUE, template_id TEXT,\n name TEXT NOT NULL, type TEXT NOT NULL, started_at TEXT NOT NULL,\n completed_at TEXT NOT NULL, duration_seconds INTEGER NOT NULL DEFAULT 0,\n rounds INTEGER, partial_reps INTEGER, effort INTEGER, feeling INTEGER,\n notes TEXT NOT NULL DEFAULT '', weight_kg REAL, distance_km REAL,\n avg_pace_sec_km INTEGER, avg_hr INTEGER, cadence_spm REAL,\n source TEXT NOT NULL DEFAULT 'val_town', source_external_id TEXT UNIQUE,\n sync_status TEXT NOT NULL DEFAULT 'pending_sync', processed_at TEXT,\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n FOREIGN KEY(template_id) REFERENCES workout_templates(id)\n )` },\n { sql: `CREATE TABLE IF NOT EXISTS laps (\n id TEXT PRIMARY KEY, session_id TEXT NOT NULL, lap_number INTEGER NOT NULL,\n elapsed_seconds INTEGER NOT NULL, reps INTEGER, note TEXT NOT NULL DEFAULT '',\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n FOREIGN KEY(session_id) REFERENCES sessions(id) ON DELETE CASCADE\n )` },\n { sql: `CREATE TABLE IF NOT EXISTS performance_entries (\n id TEXT PRIMARY KEY, session_id TEXT NOT NULL, metric_key TEXT NOT NULL,\n value REAL NOT NULL, unit TEXT NOT NULL DEFAULT 'reps',\n created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,\n FOREIGN KEY(session_id) REFERENCES sessions(id) ON DELETE CASCADE\n )` },\n { sql: `CREATE TABLE IF NOT EXISTS daily_recommendations (\n id TEXT PRIMARY KEY, for_date TEXT NOT NULL UNIQUE,\n first_choice_json TEXT NOT NULL, alternatives_json TEXT NOT NULL,\n rationale TEXT NOT NULL, based_on_session_id TEXT, model TEXT NOT NULL,\n generated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )` },\n { sql: `CREATE TABLE IF NOT EXISTS coach_state (\n key TEXT PRIMARY KEY, value_json TEXT NOT NULL,\n updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )` },\n { sql: `CREATE TABLE IF NOT EXISTS sync_state (\n key TEXT PRIMARY KEY, value TEXT NOT NULL, updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP\n )` },\n { sql: \"CREATE INDEX IF NOT EXISTS idx_sessions_started ON sessions(started_at DESC)\" },\n { sql: \"CREATE INDEX IF NOT EXISTS idx_sessions_sync ON sessions(sync_status, completed_at DESC)\" },\n { sql: \"CREATE INDEX IF NOT EXISTS idx_performance_metric ON performance_entries(metric_key, created_at DESC)\" },\n ]);\n\n await sqlite.batch(templates.map((template) => ({\n sql: `INSERT OR IGNORE INTO workout_templates\n (id, name, type, duration_seconds, rounds_target, default_weight_kg, description, tags_json, exercises_json, is_standard)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n args: [template.id, template.name, template.type, template.duration, template.rounds, template.weight,\n template.description, JSON.stringify(template.tags), JSON.stringify(template.exercises), template.isStandard],\n })));\n\n await sqlite.execute({\n sql: `INSERT OR IGNORE INTO sessions\n (id, client_id, template_id, name, type, started_at, completed_at, duration_seconds,\n rounds, effort, feeling, notes, weight_kg, source, source_external_id, sync_status, processed_at)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,\n args: [\"sheet-2026-06-14-pass-2\", \"sheet-2026-06-14-pass-2\", \"kb-complex\", \"Pass 2: KB\", \"amrap\",\n \"2026-06-14T10:00:00.000Z\", \"2026-06-14T10:20:00.000Z\", 1200, 7, 7, 4,\n \"Körde en blandning av 16 kg och 20 kg.\", 20, \"google_sheet\", \"sheet:2026-06-14:Pass 2: KB\", \"official\", \"2026-06-14T10:20:00.000Z\"],\n });\n\n await sqlite.batch([\n { sql: \"INSERT OR IGNORE INTO sync_state(key, value) VALUES ('training_primary_source', 'val_town')\" },\n { sql: \"INSERT OR IGNORE INTO sync_state(key, value) VALUES ('recommendation_owner', 'chatgpt_project_chat')\" },\n { sql: \"INSERT OR IGNORE INTO sync_state(key, value) VALUES ('legacy_sheet_ids', ?)\", args: [\"1fg2vYWJwAx3NY9Tv2usMpCO-KoeDZnsca7_XtSngJbU,1_QBlXws3llrLcIfdwR0CC33gSF_HMqFM5Pf8Kqf9PoA\"] },\n { sql: \"INSERT OR IGNORE INTO sync_state(key, value) VALUES ('legacy_sheet_mode', 'read_only')\" },\n ]);\n}\n\nfunction json(value: string | null | undefined, fallback: T): T {\n try { return value ? JSON.parse(value) : fallback; } catch { return fallback; }\n}\n\nfunction templateRow(row: Record) {\n return {\n ...row,\n tags: json(String(row.tags_json || \"\"), []),\n exercises: json(String(row.exercises_json || \"\"), []),\n is_standard: Boolean(row.is_standard),\n active: Boolean(row.active),\n };\n}\n\nfunction recommendationRow(row: Record | undefined) {\n if (!row) return null;\n return {\n ...row,\n date: row.for_date,\n first_choice: json(String(row.first_choice_json || \"\"), {}),\n alternatives: json(String(row.alternatives_json || \"\"), []),\n };\n}\n\nexport async function getBootstrapData() {\n await ensureSchema();\n const [templateResult, sessionResult, recommendationResult, frequencyResult, loadResult, skillResult] = await Promise.all([\n sqlite.execute(\"SELECT * FROM workout_templates WHERE active = 1 ORDER BY is_standard DESC, name\"),\n sqlite.execute(\"SELECT * FROM sessions ORDER BY started_at DESC LIMIT 100\"),\n sqlite.execute(\"SELECT * FROM daily_recommendations ORDER BY for_date DESC LIMIT 1\"),\n sqlite.execute(`SELECT strftime('%Y-%W', started_at) AS week, COUNT(*) AS sessions\n FROM sessions WHERE completed_at >= datetime('now', '-56 days') GROUP BY week ORDER BY week`),\n sqlite.execute(`SELECT substr(started_at, 1, 7) AS month, MAX(weight_kg) AS max_weight, COUNT(*) AS sessions\n FROM sessions WHERE weight_kg IS NOT NULL GROUP BY month ORDER BY month DESC LIMIT 6`),\n sqlite.execute(`SELECT metric_key, value, unit, created_at FROM performance_entries\n WHERE metric_key IN ('pull_ups', 'toes_to_bar') ORDER BY created_at DESC LIMIT 20`),\n ]);\n\n const sessions = sessionResult.rows as Array>;\n const frequency = frequencyResult.rows as Array>;\n const recentRuns = sessions.filter((session) => Number(session.distance_km || 0) > 0).slice(0, 8);\n const standardProgress = (templateResult.rows as Array>)\n .filter((row) => Boolean(row.is_standard))\n .map((row) => {\n const matching = sessions.filter((session) => session.template_id === row.id).slice(0, 6);\n return { template_id: row.id, name: row.name, sessions: matching };\n });\n const now = new Date();\n const currentWeek = `${now.getUTCFullYear()}-${String(Math.ceil((((now.getTime() - Date.UTC(now.getUTCFullYear(), 0, 1)) / 86400000) + new Date(Date.UTC(now.getUTCFullYear(), 0, 1)).getUTCDay() + 1) / 7)).padStart(2, \"0\")}`;\n const currentWeekSessions = Number(frequency.find((row) => row.week === currentWeek)?.sessions || 0);\n\n return {\n templates: (templateResult.rows as Array>).map(templateRow),\n sessions,\n recommendation: recommendationRow((recommendationResult.rows as Array>)[0]) || fallbackRecommendation(),\n progress: {\n sessions_this_week: currentWeekSessions,\n sessions_last_28_days: sessions.filter((session) => new Date(String(session.completed_at)).getTime() > Date.now() - 28 * 86400000).length,\n active_weeks_last_8: frequency.length,\n pending_sync: sessions.filter((session) => session.sync_status === \"pending_sync\").length,\n frequency_by_week: frequency,\n kb_load_by_month: (loadResult.rows as Array>).reverse(),\n recent_runs: recentRuns,\n standard_progress: standardProgress,\n skills: skillResult.rows,\n },\n };\n}\n\nexport async function createSession(input: SessionInput) {\n await ensureSchema();\n if (!input.name?.trim()) throw new Error(\"Passnamn saknas\");\n if (!Number.isFinite(input.duration_seconds) || input.duration_seconds < 0) throw new Error(\"Ogiltig passtid\");\n const id = crypto.randomUUID();\n const completedAt = input.completed_at || new Date().toISOString();\n const clientId = input.client_id || id;\n const statements: Array<{ sql: string; args: Array }> = [{\n sql: `INSERT INTO sessions\n (id, client_id, template_id, name, type, started_at, completed_at, duration_seconds,\n rounds, partial_reps, effort, feeling, notes, weight_kg, distance_km,\n avg_pace_sec_km, avg_hr, cadence_spm, source, sync_status)\n VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'val_town', 'pending_sync')`,\n args: [id, clientId, input.template_id || null, input.name.trim(), input.type, input.started_at, completedAt,\n Math.round(input.duration_seconds), input.rounds ?? null, input.partial_reps ?? null,\n input.effort ?? null, input.feeling ?? null, input.notes?.trim() || \"\", input.weight_kg ?? null,\n input.distance_km ?? null, input.avg_pace_sec_km ?? null, input.avg_hr ?? null, input.cadence_spm ?? null],\n }];\n (input.laps || []).forEach((lap, index) => statements.push({\n sql: \"INSERT INTO laps(id, session_id, lap_number, elapsed_seconds, reps, note) VALUES (?, ?, ?, ?, ?, ?)\",\n args: [crypto.randomUUID(), id, index + 1, Math.round(lap.elapsed_seconds), lap.reps ?? null, lap.note?.trim() || \"\"],\n }));\n (input.performance || []).forEach((entry) => statements.push({\n sql: \"INSERT INTO performance_entries(id, session_id, metric_key, value, unit) VALUES (?, ?, ?, ?, ?)\",\n args: [crypto.randomUUID(), id, entry.metric_key, entry.value, entry.unit || \"reps\"],\n }));\n await sqlite.batch(statements);\n const result = await sqlite.execute({ sql: \"SELECT * FROM sessions WHERE id = ?\", args: [id] });\n return result.rows[0];\n}\n\nexport async function getCoachContext(since?: string) {\n await ensureSchema();\n const [sessionResult, recommendation, coachStateResult] = await Promise.all([\n sqlite.execute({\n sql: `SELECT * FROM sessions\n WHERE (? IS NULL AND processed_at IS NULL) OR completed_at > ?\n ORDER BY completed_at ASC LIMIT 100`,\n args: [since || null, since || null],\n }),\n sqlite.execute(\"SELECT * FROM daily_recommendations ORDER BY for_date DESC LIMIT 1\"),\n sqlite.execute(\"SELECT value_json, updated_at FROM coach_state WHERE key = 'current' LIMIT 1\"),\n ]);\n const coachStateRow = (coachStateResult.rows as Array>)[0];\n return {\n generated_at: new Date().toISOString(),\n primary_source: \"val_town\",\n legacy_history: { mode: \"read_only_import\" },\n new_sessions: sessionResult.rows,\n coach_state: coachStateRow ? {\n ...json(String(coachStateRow.value_json || \"\"), {}),\n updated_at: coachStateRow.updated_at,\n } : null,\n latest_recommendation: recommendationRow((recommendation.rows as Array>)[0]),\n };\n}\n\nexport async function acknowledgeCoachSessions(ids: string[]) {\n await ensureSchema();\n if (!ids.length) return { updated: 0 };\n const now = new Date().toISOString();\n const results = await sqlite.batch(ids.map((id) => ({\n sql: \"UPDATE sessions SET processed_at = ? WHERE id = ?\",\n args: [now, id],\n })));\n return { updated: results.reduce((sum, result) => sum + Number(result.rowsAffected || 0), 0), processed_at: now };\n}\n\nexport function fallbackRecommendation() {\n return {\n date: new Date().toISOString().slice(0, 10),\n first_choice: { template_id: \"kb-standard-20\", title: \"Standardpass KB 20\", detail: \"20 min AMRAP · 20 kg\", intent: \"Puls och funktionell helkroppsstyrka utan viktplock.\" },\n alternatives: [\n { template_id: \"run-zone-2\", title: \"Lugn löpning\", detail: \"30 min · zon 2\" },\n { template_id: \"core-stability\", title: \"Core & stabilitet\", detail: \"15 min AMRAP\" },\n ],\n rationale: \"Ett enkelt förstaval som matchar preferensen för korta, tydliga pass. Välj zon 2 om kroppen känns sliten.\",\n model: \"fallback\",\n };\n}\n",[],[89],"d22fe9e0-412d-4047-952e-2f75e8f629b2","README.md","01a02b61-7568-77b9-9da3-bab9ae8201fe",56,["D",1787515276055],7,"# Jonas träning\n\nMobil-first personlig träningsapp i Val Town. React-frontend, Hono-API och val-scoped SQLite ligger i samma projekt. Val Town är nu primär datakälla för träningshistorik, passbibliotek, progress och dagens rekommendation.\n\n## Ansvarsfördelning\n\n- **Val Town** – source of truth för strukturerad träningsdata och appens UI/API.\n- **ChatGPT-projektchatten för Träning** – personlig coach: tolkar genomförda pass, uppdaterar prioriteringar/kalibrering och äger rekommendationslogiken.\n- **ChatGPT Scheduled Task** – kör dagligen kl. 07:00 Europe/Stockholm, läser Val Town-data + `coach_state` och skriver dagens förstaval + två alternativ till `daily_recommendations`.\n- **ChatGPT Work** – används främst för större produkt-/kodarbete. Work ska bevara datakontrakten och coachingflödet i `COACHING_WORKFLOW.md`, inte skapa en parallell coachingmodell.\n- **Intervals.icu** – automatisk aktivitetskälla från Garmin Connect. `intervals_sync.ts` körs varje timme och kan också triggas on demand från coachchatten direkt efter ett pass. Synken matchar befintliga sessioner före import för att undvika dubbletter.\n- **Google Sheets** – historisk/importkälla och backup/referens. Arken är inte längre primär source of truth och ska behandlas som read-only om inget annat uttryckligen beslutas.\n\nDet finns **ingen Val Town-cron/LLM-coach** längre. Intervals-jobbet är endast datasynk, inte coaching. Den tidigare `daily.ts` är borttagen. Dagliga rekommendationer kommer från ChatGPT Scheduled Task utanför Val Town.\n\n## V1\n\n- Home med dagens förstaval, två alternativ, övningar/reps och kort motivering.\n- Workouts med passbibliotek och övningsdetaljer.\n- Workout mode för AMRAP, EMOM, på tid och öppna pass, med timer och varv/laps.\n- History med genomförda sessioner från Val Town, inklusive importerad historik.\n- Progress för frekvens, kettlebellbelastning, löpning, standardpass och skills som pull-ups/toes-to-bar.\n- Val Town-inloggning och ägarkontroll för `jonasolof`.\n\n## Datamodell\n\n- `workout_templates` – aktiva passupplägg. **Produktionsdatabasen är auktoritativ för katalogen**; `db.ts` innehåller endast bootstrap-seed och ska inte användas som komplett lista.\n- `sessions` – genomförda pass. Nya app-pass har `source = 'val_town'`; importerad historik har `source = 'google_sheet'`.\n- `laps` – varv/minuter med elapsed time.\n- `performance_entries` – specifika progressmått, t.ex. `pull_ups` och `toes_to_bar`.\n- `coach_state` – aktuell kvalitativ coachingkontext: prioriteringar, preferenser, belastningskalibrering och relevanta begränsningar.\n- `daily_recommendations` – dagens förstaval, två alternativ, motivering, modell/ägare och session som rekommendationen bygger på.\n- `sync_state` – systemkontrakt och legacy-metadata. `training_primary_source = val_town` och `recommendation_owner = chatgpt_project_chat` är de viktiga aktuella värdena.\n\n## Coach-API\n\n`GET /api/coach/context` returnerar nya/oprocessade sessioner, aktuell `coach_state` och senaste rekommendation. Det är ett bekvämt gränssnitt för coachflöden, men ChatGPT kan också läsa SQLite direkt via Val Town-integrationen.\n\n```sql\nSELECT * FROM sessions\nWHERE processed_at IS NULL\nORDER BY completed_at ASC;\n```\n\n`processed_at` betyder **coachad/hanterad**, inte bara \"har lästs av en automation\". En morgonrekommendation ska därför inte automatiskt markera sessioner som processed.\n\n## Daglig rekommendation\n\nDen externa ChatGPT-tasken ska varje morgon:\n\n1. läsa senaste sessioner och eventuella oprocessade app-pass,\n2. läsa `performance_entries`, alla aktiva `workout_templates` och `coach_state/current`,\n3. väga träningsmix, återhämtning, aktuell kalibrering och uttalade prioriteringar,\n4. skriva/upserta dagens rad i `daily_recommendations` med `model = 'chatgpt-coach'`,\n5. endast uppdatera `coach_state` när nya passdata ger en tydlig och bestående lärdom.\n\nAppen ska presentera rekommendationen, inte själv generera en konkurrerande träningsplan.\n\n## Legacy Google Sheets\n\nHistorik och bibliotek har importerats från tidigare Google Sheets, inklusive träningsbiblioteket `1_QBlXws3llrLcIfdwR0CC33gSF_HMqFM5Pf8Kqf9PoA`. Dessa finns kvar som referens/backup, men nya produktflöden ska i första hand läsa och skriva Val Town.\n\nSe `COACHING_WORKFLOW.md` för det praktiska arbetsflödet mellan appen, coachchatten och Work.",[],[98],"4fb02f21-73bf-4594-97f0-766dc286d4f0","COACHING_WORKFLOW.md","52080aae-9e79-11f1-9940-1607ee4eb77e",57,["D",1787515281831],"# Coaching workflow\n\nDet här dokumentet beskriver hur träningsappen ska användas tillsammans med ChatGPT-projektet **Träning**. Syftet är att undvika två parallella coacher eller två olika source-of-truth-system.\n\n## Roller\n\n### Val Town\n\nVal Town är primär strukturerad datakälla och produktionsbackend.\n\nAnsvar:\n- lagra `sessions`, `laps`, `performance_entries`, `workout_templates`, `coach_state` och `daily_recommendations`,\n- driva appens UI/API,\n- ta emot nya genomförda pass från workout mode,\n- visa den rekommendation som redan finns i `daily_recommendations`.\n\nVal Town ska **inte** själv anropa en LLM för att coacha eller skapa dagens pass. Den gamla `daily.ts`-cronen är borttagen.\n\n### ChatGPT-projektchatten\n\nDen vanliga projektchatten är den personliga coachen och äger träningsresonemanget.\n\nEfter ett pass kan coachen:\n1. trigga `intervals_sync.ts` on demand för att hämta färsk Garmin/Intervals-data och därefter läsa senaste Val Town-sessionen,\n2. diskutera känsla, teknik, motivation och belastning med användaren,\n3. komplettera eller korrigera strukturerade data när det behövs,\n4. uppdatera `coach_state/current` när samtalet ger en **bestående** ny lärdom,\n5. markera sessionen `processed_at` först när den faktiskt har hanterats/coachas klart.\n\nExempel på bestående coach-state:\n- nuvarande träningsprioriteringar,\n- kalibrering av vikter/reps,\n- skills som ska byggas gradvis,\n- återkommande preferenser,\n- relevanta teknik-/belastningsbegränsningar.\n\nKortlivade kommentarer om ett enskilt pass hör normalt hemma i `sessions.notes`, inte i `coach_state`.\n\n### ChatGPT Scheduled Task\n\nEn extern ChatGPT-task kör dagligen kl. **07:00 Europe/Stockholm**.\n\nDen ska:\n1. läsa senaste `sessions`, oprocessade nya app-sessioner och `performance_entries`,\n2. läsa hela den aktiva `workout_templates`-katalogen,\n3. läsa `coach_state/current`,\n4. väga träningsmix, återhämtning och aktuella prioriteringar,\n5. skriva/upserta en rad för dagens datum i `daily_recommendations`,\n6. använda `model = 'chatgpt-coach'`,\n7. ange `based_on_session_id` när en nylig session är central för beslutet.\n\nTasken får läsa oprocessade sessioner för att göra en bättre morgonrekommendation, men ska **inte** markera dem processed bara för att de lästs.\n\n### ChatGPT Work\n\nWork används för större produkt- och kodförändringar: UI, workout mode, nya progressvyer, FIT-upload, större API-/databasrefaktorer osv.\n\nWork ska innan större ändringar läsa `README.md` och detta dokument och bevara följande kontrakt:\n- Val Town är primär träningsdatakälla.\n- Projektchatten/ChatGPT äger coachinglogiken.\n- Appen presenterar `daily_recommendations`; den genererar inte egna konkurrerande rekommendationer.\n- `coach_state` är bryggan mellan kvalitativa coachdiskussioner och strukturerad data.\n- Google Sheets är legacy/import/backup om inget annat beslutas.\n\n## Normalflöde efter ett pass\n\n```text\nWorkout mode\n ↓\nVal Town session\n ↓\nAnvändaren diskuterar passet i projektchatten\n ↓\nChatGPT läser sessionen + eventuell träningsfil\n ↓\nSession notes / performance_entries kompletteras vid behov\n ↓\nBestående lärdom → coach_state\n ↓\nNästa morgons Scheduled Task\n ↓\ndaily_recommendations\n ↓\nHome i appen\n```\n\nDet är okej om användaren inte kommer till chatten efter varje pass. Morgontasken kan ändå använda den strukturerade sessionen. Coachdiskussionen lägger framför allt till den kvalitativa tolkningen.\n\n## `coach_state/current`\n\n`coach_state` har ett JSON-fält `value_json`. Nuvarande konvention är ungefär:\n\n```json\n{\n \"priorities\": [\"...\"],\n \"preferences\": [\"...\"],\n \"calibration\": [\"...\"],\n \"constraints\": [\"...\"],\n \"updated_by\": \"chatgpt_coach\"\n}\n```\n\nHåll state kort, aktuellt och handlingsbart. Ersätt gamla slutsatser när de inte längre gäller i stället för att bara append:a historik.\n\n## `daily_recommendations`\n\nVarje datum ska ha högst en aktuell rad. Rekommendationen består av:\n- `first_choice_json` – primärt pass,\n- `alternatives_json` – två rimliga alternativ,\n- `rationale` – kort motivering utifrån senaste data,\n- `based_on_session_id` – senaste relevanta passet när tillämpligt,\n- `model` – normalt `chatgpt-coach`,\n- `generated_at`.\n\nRekommendationen ska vara konkret nog att användaren kan börja träna direkt från Home-vyn. Undvik generisk coachtext och onödig repetition av rörelsemönster från de senaste passen.\n\n## Google Sheets\n\nTidigare historik och passbibliotek har importerats från Google Sheets. De används som legacy/referens och ska inte skrivas till automatiskt från appen eller Work utan ett nytt uttryckligt beslut.\n",[],[107],"bf91df80-1153-4615-a8e1-a5870d81a0cb","CHANGELOG.md","01a02b61-7568-77b9-9da3-b61964261763",44,["D",1787437950110],2,"# Changelog\n\n## 2026-08-23 — Coach ownership and Val Town source of truth\n\n- Val Town is now the primary structured training data source.\n- Added persistent `coach_state` for current priorities, preferences, calibration and constraints.\n- ChatGPT project chat owns coaching logic; a ChatGPT Scheduled Task writes `daily_recommendations` each morning.\n- Removed the unused Val Town LLM/cron file `daily.ts` and its dead recommendation helper functions.\n- `GET /api/coach/context` now includes `coach_state` and identifies Val Town as the primary source.\n- Added `COACHING_WORKFLOW.md` so future Work sessions preserve the coaching/data contract.\n\n---\n\n## Upstream template history\n\n### 2026-07-30 — Immutable asset caching\n\nFrontend files are now served at version-addressed URLs\n(`/__immutable/45/frontend/index.tsx`) with\n`Cache-Control: public, max-age=31536000, immutable`, so repeat visits load the\nwhole client graph from the browser cache — roughly 3× faster (~665ms → ~157ms\nmeasured). Publishing your val bumps its version, which changes every URL, so\ncaches invalidate automatically. Old-version URLs return 404 after a publish\n(like Next.js build assets); a reload picks up the new version.\n\nWhat changed in the template:\n\n- `frontend/index.html` → [`frontend/root.tsx`](/frontend/root.tsx): the HTML\n shell is now JSX and stamps asset URLs with `immutableFileUrl()`\n- [`index.ts`](/index.ts): serves `Root()` at `/`, and one\n `app.get(\"/__immutable/*\", (c) => serveImmutableFile(c.req.path))` route\n serves all frontend files\n\n### Upgrading an existing remix\n\nYour copy of this template doesn't update automatically. Two options:\n\n**Minimal (most of the benefit):** point both routes at `serveImmutableFile` —\nyour existing bare URLs redirect into versioned space, the new route serves\nthem, and the whole module graph gets cached (costs one redirect per page view):\n\n```ts\nimport { serveImmutableFile } from \"https://esm.town/v/std/utils/index.ts\";\n\napp.get(\"/__immutable/*\", (c) => serveImmutableFile(c.req.path));\napp.get(\"/frontend/**/*\", (c) => serveImmutableFile(c.req.path)); // was serveFile\n```\n\n**Full upgrade (also removes the redirect):** three self-contained steps —\nno other files needed. Adjust the ``, favicon, and body of `Root()` to\nmatch your app's existing `index.html`, and stamp anything else your shell\nreferences directly (images, css) with `immutableFileUrl()`.\n\n1. Create `frontend/root.tsx`:\n\n```tsx\n/** @jsxImportSource npm:hono@4/jsx */\nimport { raw } from \"npm:hono@4/html\";\nimport { immutableFileUrl } from \"https://esm.town/v/std/utils/index.ts\";\n\n// The HTML shell for the app. It is never cached, so immutableFileUrl can stamp\n// the current val version into each asset URL — the browser caches those\n// immutably, and publishing the val changes the URLs, refetching exactly once.\nexport function Root() {\n return (\n <>\n {raw(\"<!DOCTYPE html>\")}\n <html lang=\"en\">\n <head>\n <meta charSet=\"UTF-8\" />\n <meta name=\"viewport\" content=\"width=device-width, initial-scale=1.0\" />\n <title>React Hono Val Town Starter\n