[{"_1":2,"_181":-5,"_182":-5},"loaderData",{"_3":4,"_93":94},"root",{"_5":6,"_36":37,"_66":67},"ctx",{"_7":8,"_23":-5,"_24":25,"_26":27,"_28":20,"_29":20,"_30":20,"_31":-5,"_32":33,"_34":20,"_35":20},"auth",{"_9":-5,"_10":-5,"_11":-5,"_12":-5,"_13":-5,"_14":-5,"_15":-5,"_16":-5,"_17":-5,"_18":-5,"_19":20,"_21":22},"sessionClaims","sessionId","sessionStatus","userId","actor","orgId","orgRole","orgSlug","orgPermissions","factorVerificationAge","isAuthenticated",false,"tokenType","session_token","user","preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesChatPanelState","notificationCount",0,"flagDenols","isTeamOrgMember","ENV",{"_38":39,"_40":41,"_42":43,"_44":45,"_46":47,"_48":49,"_50":51,"_52":53,"_54":55,"_56":57,"_58":59,"_60":61,"_62":63,"_64":65},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","VAL_RUNTIME_NAME","prod","RENDER_GIT_COMMIT","08df2158b567c8dc30dd332965dd8203fb644318","RENDER_GIT_BRANCH","main","CLERK_PUBLISHABLE_KEY","pk_live_Y2xlcmsudmFsLnRvd24k","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","API_URL","https://api.val.town","clerkState",{"_68":69},"__internal_clerk_state",{"_70":71,"_72":47,"_73":74,"_75":74,"_76":20,"_77":74,"_78":74,"_79":74,"_80":74,"_81":82,"_91":20,"_92":20},"__clerk_ssr_state",{"_21":22,"_9":-5,"_10":-5,"_11":-5,"_12":-5,"_13":-5,"_14":-5,"_15":-5,"_16":-5,"_17":-5,"_18":-5,"_19":20},"__publishableKey","__proxyUrl","","__domain","__isSatellite","__signInUrl","__signUpUrl","__afterSignInUrl","__afterSignUpUrl","__clerk_debug",{"_83":20,"_19":20,"_84":74,"_85":86,"_87":74,"_88":47,"_89":20,"_90":74},"isSignedIn","proxyUrl","reason","session-token-and-uat-missing","message","publishableKey","isSatellite","domain","__telemetryDisabled","__telemetryDebug","routes/embed.x.$handle.$name.$",{"_95":96,"_97":98,"_99":100,"_101":102,"_155":156,"_178":179,"_180":-5},"handle","kamenxrider","name","readwise-mastra","path","lib/agent.ts","project",{"_103":104,"_97":98,"_105":106,"_107":-5,"_108":20,"_109":110,"_111":-5,"_112":-5,"_12":113,"_114":-5,"_115":20,"_116":-5,"_117":-5,"_118":33,"_14":119,"_120":121,"_122":123,"_124":125,"_126":125,"_127":128,"_129":128,"_130":128,"_131":125,"_132":125,"_23":133,"_136":137,"_141":142,"_154":20},"id","cbd319ae-ffd3-11f0-bcd8-42dde27851f2","createdAt",["D",1769994557115],"deletedAt","legacyEnvVars","privacy","public","description","imageUrl","d4de366f-8281-484e-8856-3824b6bdfb0b","pinnedFileId","pinPreview","pinMigrated","screenshotVersion","likeCount","9bb2e2b6-8aa1-4b6a-b4ef-ad7adab2b868","tursoName","cbd319aeffd311f0bcd842dde27851f2","tursoHostname","cbd319aeffd311f0bcd842dde27851f2-val-town.aws-us-east-2.turso.io","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm",{"_95":96,"_134":135},"profileImageUrl","https://img.clerk.com/eyJ0eXBlIjoicHJveHkiLCJzcmMiOiJodHRwczovL2ltYWdlcy5jbGVyay5kZXYvb2F1dGhfZ2l0aHViL2ltZ18ybFJkNm8zSzFzbGE5RzhsTUt3OGJMZFdGZFkifQ","org",{"_95":96,"_138":135,"_12":113,"_139":140},"avatar","tier","pro","branch",{"_143":144,"_97":45,"_103":145,"_146":104,"_147":-5,"_148":-5,"_149":-5,"_150":-5,"_151":-5,"_152":98,"_153":96},"version",51,"cbd4b750-ffd3-11f0-bcd8-42dde27851f2","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","isOwned","file",{"_103":157,"_97":158,"_159":160,"_161":162,"_163":145,"_164":165,"_166":167,"_105":168,"_143":169,"_170":171,"_172":173,"_174":20,"_175":-5,"_99":176},"d72d97ee-ffd7-11f0-9a77-42dde27851f2","agent.ts","fileId","fe664cdc-ffd4-11f0-97f3-42dde27851f2","parentId","99160818-ffd4-11f0-97f3-42dde27851f2","branchId","branchVersion",44,"latest",true,["D",1769996294152],4,"fileType","script","content","/**\n * AI agent logic + structured plan schema\n * Uses Pollinations as the LLM provider\n * \n * Note: We use a simplified agent implementation that calls Pollinations directly\n * rather than the full Mastra Agent class, to avoid zod version compatibility issues.\n * The tool definitions are kept for future Mastra integration when stable.\n */\n\nimport { z } from \"npm:zod@3.23.8\";\nimport { sqlite } from \"https://esm.town/v/std/sqlite@14-main/main.ts\";\nimport { generateJSON, generateChat, type ChatMessage } from \"./llm_pollinations.ts\";\nimport { updateDoc as readwiseUpdateDoc, type UpdateDocParams, RateLimitError } from \"./readwise.ts\";\nimport { mapRows, upsertDoc, setDocTags, getDocTags, type Doc } from \"./db.ts\";\n\n// ============ Plan Schema ============\n\nexport const PatchSchema = z.object({\n location: z.enum([\"new\", \"later\", \"shortlist\", \"archive\", \"feed\"]).optional(),\n tagsAdd: z.array(z.string()).optional(),\n tagsRemove: z.array(z.string()).optional(),\n notes: z.string().optional(),\n summary: z.string().optional(),\n title: z.string().optional(),\n});\n\nexport const PlanItemSchema = z.object({\n docId: z.string(),\n patch: PatchSchema,\n rationale: z.string(),\n});\n\nexport const PlanSchema = z.object({\n items: z.array(PlanItemSchema),\n});\n\nexport type Patch = z.infer;\nexport type PlanItem = z.infer;\nexport type Plan = z.infer;\n\n// ============ Tool Functions (for future Mastra integration) ============\n\n/**\n * Search the local SQLite index\n */\nexport async function searchIndex(params: {\n query?: string;\n location?: string;\n category?: string;\n tag?: string;\n limit?: number;\n}): Promise<{ docs: Array<{\n id: string;\n title: string | null;\n author: string | null;\n site_name: string | null;\n category: string | null;\n location: string | null;\n reading_progress: number | null;\n}>; total: number }> {\n const { query, location, category, tag, limit = 20 } = params;\n const conditions: string[] = [];\n const args: (string | number)[] = [];\n\n if (query) {\n conditions.push(\"(title LIKE ? OR author LIKE ? OR site_name LIKE ?)\");\n const likePattern = `%${query}%`;\n args.push(likePattern, likePattern, likePattern);\n }\n if (location) {\n conditions.push(\"location = ?\");\n args.push(location);\n }\n if (category) {\n conditions.push(\"category = ?\");\n args.push(category);\n }\n\n let sql = `SELECT id, title, author, site_name, category, location, reading_progress FROM docs`;\n\n if (tag) {\n sql = `SELECT d.id, d.title, d.author, d.site_name, d.category, d.location, d.reading_progress\n FROM docs d\n JOIN doc_tags dt ON d.id = dt.doc_id\n WHERE dt.tag_key = ?`;\n args.unshift(tag);\n if (conditions.length > 0) {\n sql += ` AND ${conditions.join(\" AND \")}`;\n }\n } else if (conditions.length > 0) {\n sql += ` WHERE ${conditions.join(\" AND \")}`;\n }\n\n sql += ` ORDER BY updated_at DESC LIMIT ?`;\n args.push(limit);\n\n const result = await sqlite.execute({ sql, args });\n const docs = mapRows<{\n id: string;\n title: string | null;\n author: string | null;\n site_name: string | null;\n category: string | null;\n location: string | null;\n reading_progress: number | null;\n }>(result);\n\n // Get total count\n let countSql = \"SELECT COUNT(*) as count FROM docs\";\n const countArgs: (string | number)[] = [];\n\n if (tag) {\n countSql = `SELECT COUNT(*) as count FROM docs d JOIN doc_tags dt ON d.id = dt.doc_id WHERE dt.tag_key = ?`;\n countArgs.push(tag);\n }\n\n const countResult = await sqlite.execute({ sql: countSql, args: countArgs });\n const total = (countResult.rows[0]?.[0] as number) || 0;\n\n return { docs, total };\n}\n\n/**\n * Get full document details\n */\nexport async function getDoc(docId: string): Promise<{\n id: string;\n title: string | null;\n author: string | null;\n site_name: string | null;\n source_url: string | null;\n url: string | null;\n category: string | null;\n location: string | null;\n word_count: number | null;\n reading_progress: number | null;\n notes: string | null;\n summary: string | null;\n saved_at: string | null;\n tags: string[];\n} | null> {\n const result = await sqlite.execute({\n sql: \"SELECT * FROM docs WHERE id = ?\",\n args: [docId],\n });\n\n if (result.rows.length === 0) {\n return null;\n }\n\n const doc = mapRows(result)[0];\n const tags = await getDocTags(docId);\n\n return {\n id: doc.id,\n title: doc.title,\n author: doc.author,\n site_name: doc.site_name,\n source_url: doc.source_url,\n url: doc.url,\n category: doc.category,\n location: doc.location,\n word_count: doc.word_count,\n reading_progress: doc.reading_progress,\n notes: doc.notes,\n summary: doc.summary,\n saved_at: doc.saved_at,\n tags,\n };\n}\n\n/**\n * Validate a patch\n */\nexport function validatePatch(patch: Patch): { valid: boolean; errors: string[] } {\n const errors: string[] = [];\n\n // Validate location if present\n const validLocations = [\"new\", \"later\", \"shortlist\", \"archive\", \"feed\"];\n if (patch.location && !validLocations.includes(patch.location)) {\n errors.push(`Invalid location: ${patch.location}. Must be one of: ${validLocations.join(\", \")}`);\n }\n\n // Validate tags are strings\n if (patch.tagsAdd && !patch.tagsAdd.every((t) => typeof t === \"string\")) {\n errors.push(\"tagsAdd must be an array of strings\");\n }\n if (patch.tagsRemove && !patch.tagsRemove.every((t) => typeof t === \"string\")) {\n errors.push(\"tagsRemove must be an array of strings\");\n }\n\n return { valid: errors.length === 0, errors };\n}\n\n/**\n * Apply a patch to Readwise and local DB\n */\nexport async function applyPatch(\n docId: string,\n patch: Patch\n): Promise<{ success: boolean; error?: string }> {\n try {\n // Build Readwise update params\n const updateParams: UpdateDocParams = {};\n\n if (patch.location) updateParams.location = patch.location;\n if (patch.title) updateParams.title = patch.title;\n if (patch.summary) updateParams.summary = patch.summary;\n if (patch.notes) updateParams.notes = patch.notes;\n\n // Handle tags (get current, add/remove, set new)\n if (patch.tagsAdd || patch.tagsRemove) {\n const currentTags = await getDocTags(docId);\n let newTags = [...currentTags];\n\n if (patch.tagsRemove) {\n newTags = newTags.filter((t) => !patch.tagsRemove!.includes(t));\n }\n if (patch.tagsAdd) {\n for (const tag of patch.tagsAdd) {\n if (!newTags.includes(tag)) {\n newTags.push(tag);\n }\n }\n }\n\n updateParams.tags = newTags;\n }\n\n // Update Readwise\n const updatedDoc = await readwiseUpdateDoc(docId, updateParams);\n\n // Update local DB\n await upsertDoc({\n id: docId,\n location: updatedDoc.location,\n title: updatedDoc.title,\n summary: updatedDoc.summary,\n notes: updatedDoc.notes,\n updated_at: updatedDoc.updated_at,\n });\n\n // Update local tags\n if (updateParams.tags) {\n await setDocTags(docId, updateParams.tags);\n }\n\n return { success: true };\n } catch (err) {\n return {\n success: false,\n error: err instanceof Error ? err.message : String(err),\n };\n }\n}\n\n// ============ Plan Generation ============\n\n/**\n * Generate a plan using the LLM\n * Returns a structured plan that can be reviewed before applying\n */\nexport async function generatePlan(\n goal: string,\n scope: {\n location?: string;\n category?: string;\n tag?: string;\n limit?: number;\n }\n): Promise<{ plan: Plan; docsAnalyzed: number }> {\n // First, fetch documents matching the scope\n const conditions: string[] = [];\n const args: (string | number)[] = [];\n\n if (scope.location) {\n conditions.push(\"location = ?\");\n args.push(scope.location);\n }\n if (scope.category) {\n conditions.push(\"category = ?\");\n args.push(scope.category);\n }\n\n let sql = `SELECT id, title, author, site_name, category, location, word_count, reading_progress, summary, notes\n FROM docs`;\n\n if (scope.tag) {\n sql = `SELECT d.id, d.title, d.author, d.site_name, d.category, d.location, d.word_count, d.reading_progress, d.summary, d.notes\n FROM docs d\n JOIN doc_tags dt ON d.id = dt.doc_id\n WHERE dt.tag_key = ?`;\n args.unshift(scope.tag);\n if (conditions.length > 0) {\n sql += ` AND ${conditions.join(\" AND \")}`;\n }\n } else if (conditions.length > 0) {\n sql += ` WHERE ${conditions.join(\" AND \")}`;\n }\n\n sql += ` ORDER BY updated_at DESC LIMIT ?`;\n args.push(scope.limit || 50);\n\n const result = await sqlite.execute({ sql, args });\n const docs = mapRows<{\n id: string;\n title: string | null;\n author: string | null;\n site_name: string | null;\n category: string | null;\n location: string | null;\n word_count: number | null;\n reading_progress: number | null;\n summary: string | null;\n notes: string | null;\n }>(result);\n\n if (docs.length === 0) {\n return { plan: { items: [] }, docsAnalyzed: 0 };\n }\n\n // Build prompt for the LLM\n const docsDescription = docs\n .map((d, i) => {\n return `${i + 1}. [${d.id}] \"${d.title || \"Untitled\"}\" by ${d.author || \"Unknown\"}\n - Site: ${d.site_name || \"N/A\"}, Category: ${d.category}, Location: ${d.location}\n - Words: ${d.word_count || \"?\"}, Progress: ${Math.round((d.reading_progress || 0) * 100)}%\n - Summary: ${d.summary?.slice(0, 100) || \"N/A\"}`;\n })\n .join(\"\\n\\n\");\n\n const prompt = `You are helping organize a Readwise Reader library.\n\nGOAL: ${goal}\n\nDOCUMENTS TO ANALYZE:\n${docsDescription}\n\nBased on the goal, create a plan of patches to apply to these documents.\nEach patch can include:\n- location: \"new\", \"later\", \"shortlist\", \"archive\", or \"feed\"\n- tagsAdd: array of tag names to add\n- tagsRemove: array of tag names to remove\n- notes: text to set as notes\n- summary: text to set as summary\n- title: new title\n\nOnly include patches for documents that need changes. Provide a brief rationale for each.\n\nRespond with a JSON object in this exact format:\n{\n \"items\": [\n {\n \"docId\": \"document-id\",\n \"patch\": { \"location\": \"archive\", \"tagsAdd\": [\"read\"] },\n \"rationale\": \"Brief explanation\"\n }\n ]\n}`;\n\n const systemPrompt = `You are a document organization assistant. Analyze documents and create actionable plans.\nAlways respond with valid JSON matching the requested schema. Be concise in rationales.`;\n\n // Use complexTask: true for plan generation to get better model\n const { data, modelUsed } = await generateJSON(prompt, {\n systemPrompt,\n temperature: 0.3,\n maxTokens: 4096,\n complexTask: true, // Use stronger model for planning\n });\n\n console.log(`[agent] Plan generated using model: ${modelUsed}`);\n\n // Validate the plan\n const parsed = PlanSchema.safeParse(data);\n if (!parsed.success) {\n console.error(\"Plan validation failed:\", parsed.error);\n return { plan: { items: [] }, docsAnalyzed: docs.length };\n }\n\n return { plan: parsed.data, docsAnalyzed: docs.length };\n}\n\n/**\n * Apply a plan (execute all patches)\n */\nexport async function applyPlan(\n plan: Plan,\n onProgress?: (completed: number, total: number, error?: string) => void\n): Promise<{ applied: number; errors: string[] }> {\n let applied = 0;\n const errors: string[] = [];\n\n for (let i = 0; i < plan.items.length; i++) {\n const item = plan.items[i];\n\n try {\n // Build Readwise update params\n const updateParams: UpdateDocParams = {};\n\n if (item.patch.location) updateParams.location = item.patch.location;\n if (item.patch.title) updateParams.title = item.patch.title;\n if (item.patch.summary) updateParams.summary = item.patch.summary;\n if (item.patch.notes) updateParams.notes = item.patch.notes;\n\n // Handle tags\n if (item.patch.tagsAdd || item.patch.tagsRemove) {\n const currentTags = await getDocTags(item.docId);\n let newTags = [...currentTags];\n\n if (item.patch.tagsRemove) {\n newTags = newTags.filter((t) => !item.patch.tagsRemove!.includes(t));\n }\n if (item.patch.tagsAdd) {\n for (const tag of item.patch.tagsAdd) {\n if (!newTags.includes(tag)) {\n newTags.push(tag);\n }\n }\n }\n\n updateParams.tags = newTags;\n }\n\n // Update Readwise\n const updatedDoc = await readwiseUpdateDoc(item.docId, updateParams);\n\n // Update local DB\n await upsertDoc({\n id: item.docId,\n location: updatedDoc.location,\n title: updatedDoc.title,\n summary: updatedDoc.summary,\n notes: updatedDoc.notes,\n updated_at: updatedDoc.updated_at,\n });\n\n if (updateParams.tags) {\n await setDocTags(item.docId, updateParams.tags);\n }\n\n applied++;\n onProgress?.(applied, plan.items.length);\n\n // Small delay between updates to be nice to the API\n if (i < plan.items.length - 1) {\n await new Promise((resolve) => setTimeout(resolve, 200));\n }\n } catch (err) {\n if (err instanceof RateLimitError) {\n console.log(`Rate limited, waiting ${err.retryAfter} seconds...`);\n await new Promise((resolve) => setTimeout(resolve, err.retryAfter * 1000));\n // Retry this item\n i--;\n continue;\n }\n\n const errorMsg = `Failed to apply patch to ${item.docId}: ${err instanceof Error ? err.message : String(err)}`;\n errors.push(errorMsg);\n onProgress?.(applied, plan.items.length, errorMsg);\n }\n }\n\n return { applied, errors };\n}\n\n// ============ Multi-step Agent (simplified) ============\n\n/**\n * Run an agent conversation with tool use\n * This is a simplified implementation that can be extended to use Mastra Agent.generate() later\n */\nexport async function runAgent(\n goal: string,\n maxSteps: number = 5\n): Promise<{ response: string; toolCalls: Array<{ tool: string; result: unknown }> }> {\n const toolCalls: Array<{ tool: string; result: unknown }> = [];\n const messages: ChatMessage[] = [\n {\n role: \"system\",\n content: `You are a helpful assistant for organizing a Readwise Reader library.\nYou have access to tools to search documents, get document details, and apply patches.\nWhen the user asks you to organize documents, analyze their request and provide a helpful response.\n\nAvailable tools (describe what you want to do and I'll execute it):\n- searchIndex: Search documents by query, location, category, or tag\n- getDoc: Get full details of a specific document\n- generatePlan: Generate an AI plan for a given goal\n\nDescribe your reasoning and what actions you want to take.`,\n },\n {\n role: \"user\",\n content: goal,\n },\n ];\n\n for (let step = 0; step < maxSteps; step++) {\n // Use cheaper model for interactive agent steps\n const { text, modelUsed } = await generateChat(messages, {\n temperature: 0.7,\n complexTask: false, // Use cheaper model for simple interactions\n });\n console.log(`[agent] Step ${step + 1} using model: ${modelUsed}`);\n\n messages.push({ role: \"assistant\", content: text });\n\n // Check if the response indicates a tool call (simple pattern matching)\n // In a full implementation, this would use structured tool calling\n if (text.toLowerCase().includes(\"search for\") || text.toLowerCase().includes(\"let me search\")) {\n // Execute search\n const searchResult = await searchIndex({ limit: 10 });\n toolCalls.push({ tool: \"searchIndex\", result: searchResult });\n messages.push({\n role: \"user\",\n content: `Search results: ${JSON.stringify(searchResult.docs.slice(0, 5))}`,\n });\n continue;\n }\n\n // If no tool call detected, we're done\n break;\n }\n\n const lastMessage = messages[messages.length - 1];\n return {\n response: typeof lastMessage.content === \"string\" ? lastMessage.content : JSON.stringify(lastMessage.content),\n toolCalls,\n };\n}\n","deleted","label",[177,158],"lib","permalink","https://www.val.town/x/kamenxrider/readwise-mastra/code/lib/agent.ts","extension","actionData","errors"]