llmist 0.1.6 → 0.2.1

package/dist/cli.cjs

@@ -86,17 +86,17 @@ var init_model_shortcuts = __esm({
       "gpt5-mini": "openai:gpt-5-mini",
       "gpt5-nano": "openai:gpt-5-nano",
       // Anthropic aliases
-      sonnet: "anthropic:claude-3-5-sonnet-latest",
-      "claude-sonnet": "anthropic:claude-3-5-sonnet-latest",
-      haiku: "anthropic:claude-3-5-haiku-latest",
-      "claude-haiku": "anthropic:claude-3-5-haiku-latest",
-      opus: "anthropic:claude-3-opus-latest",
-      "claude-opus": "anthropic:claude-3-opus-latest",
+      sonnet: "anthropic:claude-sonnet-4-5",
+      "claude-sonnet": "anthropic:claude-sonnet-4-5",
+      haiku: "anthropic:claude-haiku-4-5",
+      "claude-haiku": "anthropic:claude-haiku-4-5",
+      opus: "anthropic:claude-opus-4-5",
+      "claude-opus": "anthropic:claude-opus-4-5",
       // Gemini aliases
       flash: "gemini:gemini-2.0-flash",
       "gemini-flash": "gemini:gemini-2.0-flash",
-      "gemini-pro": "gemini:gemini-2.0-pro",
-      pro: "gemini:gemini-2.0-pro"
+      "gemini-pro": "gemini:gemini-2.5-pro",
+      pro: "gemini:gemini-2.5-pro"
     };
     KNOWN_MODEL_PATTERNS = [
       /^gpt-?\d/i,
@@ -2070,30 +2070,6 @@ var init_anthropic_models = __esm({
   "src/providers/anthropic-models.ts"() {
     "use strict";
     ANTHROPIC_MODELS = [
-      {
-        provider: "anthropic",
-        modelId: "claude-opus-4-5-20251124",
-        displayName: "Claude Opus 4.5",
-        contextWindow: 2e5,
-        maxOutputTokens: 64e3,
-        pricing: {
-          input: 5,
-          output: 25,
-          cachedInput: 0.5
-        },
-        knowledgeCutoff: "2025-03",
-        features: {
-          streaming: true,
-          functionCalling: true,
-          vision: true,
-          reasoning: true
-        },
-        metadata: {
-          family: "Claude 4",
-          releaseDate: "2025-11-24",
-          notes: "Most powerful model. 80.9% SWE-bench Verified, 66.3% OSWorld. Best for coding and computer use."
-        }
-      },
       {
         provider: "anthropic",
         modelId: "claude-sonnet-4-5-20250929",
@@ -2282,6 +2258,79 @@ var init_anthropic_models = __esm({
           releaseDate: "2024-03-07",
           notes: "Legacy model - upgrade to Haiku 4.5 for better performance"
         }
+      },
+      // Modern aliases (recommended by Anthropic)
+      {
+        provider: "anthropic",
+        modelId: "claude-haiku-4-5",
+        displayName: "Claude Haiku 4.5",
+        contextWindow: 2e5,
+        maxOutputTokens: 64e3,
+        pricing: {
+          input: 1,
+          output: 5,
+          cachedInput: 0.1
+        },
+        knowledgeCutoff: "2025-02",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true
+        },
+        metadata: {
+          family: "Claude 4",
+          releaseDate: "2025-10-01",
+          notes: "Alias for claude-haiku-4-5-20251001. Fastest model with near-frontier intelligence."
+        }
+      },
+      {
+        provider: "anthropic",
+        modelId: "claude-sonnet-4-5",
+        displayName: "Claude Sonnet 4.5",
+        contextWindow: 2e5,
+        maxOutputTokens: 64e3,
+        pricing: {
+          input: 3,
+          output: 15,
+          cachedInput: 0.3
+        },
+        knowledgeCutoff: "2025-01",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true
+        },
+        metadata: {
+          family: "Claude 4",
+          releaseDate: "2025-09-29",
+          notes: "Alias for claude-sonnet-4-5-20250929. Smartest model for complex agents and coding."
+        }
+      },
+      {
+        provider: "anthropic",
+        modelId: "claude-opus-4-5",
+        displayName: "Claude Opus 4.5",
+        contextWindow: 2e5,
+        maxOutputTokens: 64e3,
+        pricing: {
+          input: 5,
+          output: 25,
+          cachedInput: 0.5
+        },
+        knowledgeCutoff: "2025-03",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true
+        },
+        metadata: {
+          family: "Claude 4",
+          releaseDate: "2025-11-24",
+          notes: "Alias for claude-opus-4-5-20251124. Most powerful model for coding and computer use."
+        }
       }
     ];
   }
@@ -4216,7 +4265,8 @@ var CLI_NAME = "llmist";
 var CLI_DESCRIPTION = "Command line utilities for llmist agents and direct LLM access.";
 var COMMANDS = {
   complete: "complete",
-  agent: "agent"
+  agent: "agent",
+  models: "models"
 };
 var LOG_LEVELS = ["silly", "trace", "debug", "info", "warn", "error", "fatal"];
 var DEFAULT_MODEL = "openai:gpt-5-nano";
@@ -4234,7 +4284,7 @@ var OPTION_FLAGS = {
   noBuiltins: "--no-builtins"
 };
 var OPTION_DESCRIPTIONS = {
-  model: "Model identifier, e.g. openai:gpt-5-nano or anthropic:claude-3-5-sonnet-latest.",
+  model: "Model identifier, e.g. openai:gpt-5-nano or anthropic:claude-sonnet-4-5.",
   systemPrompt: "Optional system prompt prepended to the conversation.",
   temperature: "Sampling temperature between 0 and 2.",
   maxTokens: "Maximum number of output tokens requested from the model.",
@@ -4253,7 +4303,7 @@ var import_commander3 = require("commander");
 // package.json
 var package_default = {
   name: "llmist",
-  version: "0.1.6",
+  version: "0.2.0",
   description: "Universal TypeScript LLM client with streaming-first agent framework. Works with any model - no structured outputs or native tool calling required. Implements its own flexible grammar for function calling.",
   type: "module",
   main: "dist/index.cjs",
@@ -4295,7 +4345,8 @@ var package_default = {
     "test:e2e:watch": "bun test src/e2e --watch --timeout 60000",
     "test:all": "bun run test && bun run test:e2e",
     clean: "rimraf dist",
-    prepare: "node scripts/install-hooks.js || true"
+    prepare: "node scripts/install-hooks.js || true",
+    "release:dry": "bunx semantic-release --dry-run"
   },
   bin: {
     llmist: "dist/cli.js"
@@ -4344,11 +4395,16 @@ var package_default = {
   },
   devDependencies: {
     "@biomejs/biome": "^2.3.2",
+    "@commitlint/cli": "^20.1.0",
+    "@commitlint/config-conventional": "^20.0.0",
+    "@semantic-release/changelog": "^6.0.3",
+    "@semantic-release/git": "^10.0.1",
     "@types/js-yaml": "^4.0.9",
     "@types/node": "^20.12.7",
     "bun-types": "^1.3.2",
     dotenv: "^17.2.3",
     rimraf: "^5.0.5",
+    "semantic-release": "^25.0.2",
     tsup: "^8.3.5",
     typescript: "^5.4.5"
   }
@@ -5279,9 +5335,169 @@ function registerCompleteCommand(program, env) {
   );
 }
 
+// src/cli/models-command.ts
+var import_chalk3 = __toESM(require("chalk"), 1);
+init_model_shortcuts();
+async function handleModelsCommand(options, env) {
+  const client = env.createClient();
+  const models = client.modelRegistry.listModels(options.provider);
+  if (options.format === "json") {
+    renderJSON(models, env.stdout);
+  } else {
+    renderTable(models, options.verbose || false, env.stdout);
+  }
+}
+function renderTable(models, verbose, stream2) {
+  const grouped = /* @__PURE__ */ new Map();
+  for (const model of models) {
+    const provider = model.provider;
+    if (!grouped.has(provider)) {
+      grouped.set(provider, []);
+    }
+    grouped.get(provider).push(model);
+  }
+  stream2.write(import_chalk3.default.bold.cyan("\nAvailable Models\n"));
+  stream2.write(import_chalk3.default.cyan("=".repeat(80)) + "\n\n");
+  const providers = Array.from(grouped.keys()).sort();
+  for (const provider of providers) {
+    const providerModels = grouped.get(provider);
+    const providerName = provider.charAt(0).toUpperCase() + provider.slice(1);
+    stream2.write(import_chalk3.default.bold.yellow(`${providerName} Models
+`));
+    if (verbose) {
+      renderVerboseTable(providerModels, stream2);
+    } else {
+      renderCompactTable(providerModels, stream2);
+    }
+    stream2.write("\n");
+  }
+  stream2.write(import_chalk3.default.bold.magenta("Model Shortcuts\n"));
+  stream2.write(import_chalk3.default.dim("\u2500".repeat(80)) + "\n");
+  const shortcuts = Object.entries(MODEL_ALIASES).sort((a, b) => a[0].localeCompare(b[0]));
+  for (const [shortcut, fullName] of shortcuts) {
+    stream2.write(import_chalk3.default.cyan(`  ${shortcut.padEnd(15)}`) + import_chalk3.default.dim(" \u2192 ") + import_chalk3.default.white(fullName) + "\n");
+  }
+  stream2.write("\n");
+}
+function renderCompactTable(models, stream2) {
+  const idWidth = 25;
+  const nameWidth = 22;
+  const contextWidth = 13;
+  const inputWidth = 10;
+  const outputWidth = 10;
+  stream2.write(import_chalk3.default.dim("\u2500".repeat(idWidth + nameWidth + contextWidth + inputWidth + outputWidth + 8)) + "\n");
+  stream2.write(
+    import_chalk3.default.bold(
+      "Model ID".padEnd(idWidth) + "  " + "Display Name".padEnd(nameWidth) + "  " + "Context".padEnd(contextWidth) + "  " + "Input".padEnd(inputWidth) + "  " + "Output".padEnd(outputWidth)
+    ) + "\n"
+  );
+  stream2.write(import_chalk3.default.dim("\u2500".repeat(idWidth + nameWidth + contextWidth + inputWidth + outputWidth + 8)) + "\n");
+  for (const model of models) {
+    const contextFormatted = formatTokens(model.contextWindow);
+    const inputPrice = `$${model.pricing.input.toFixed(2)}`;
+    const outputPrice = `$${model.pricing.output.toFixed(2)}`;
+    stream2.write(
+      import_chalk3.default.green(model.modelId.padEnd(idWidth)) + "  " + import_chalk3.default.white(model.displayName.padEnd(nameWidth)) + "  " + import_chalk3.default.yellow(contextFormatted.padEnd(contextWidth)) + "  " + import_chalk3.default.cyan(inputPrice.padEnd(inputWidth)) + "  " + import_chalk3.default.cyan(outputPrice.padEnd(outputWidth)) + "\n"
+    );
+  }
+  stream2.write(import_chalk3.default.dim("\u2500".repeat(idWidth + nameWidth + contextWidth + inputWidth + outputWidth + 8)) + "\n");
+  stream2.write(import_chalk3.default.dim(`  * Prices are per 1M tokens
+`));
+}
+function renderVerboseTable(models, stream2) {
+  for (const model of models) {
+    stream2.write(import_chalk3.default.bold.green(`
+  ${model.modelId}
+`));
+    stream2.write(import_chalk3.default.dim("  " + "\u2500".repeat(60)) + "\n");
+    stream2.write(`  ${import_chalk3.default.dim("Name:")}         ${import_chalk3.default.white(model.displayName)}
+`);
+    stream2.write(`  ${import_chalk3.default.dim("Context:")}      ${import_chalk3.default.yellow(formatTokens(model.contextWindow))}
+`);
+    stream2.write(`  ${import_chalk3.default.dim("Max Output:")}   ${import_chalk3.default.yellow(formatTokens(model.maxOutputTokens))}
+`);
+    stream2.write(`  ${import_chalk3.default.dim("Pricing:")}      ${import_chalk3.default.cyan(`$${model.pricing.input.toFixed(2)} input`)} ${import_chalk3.default.dim("/")} ${import_chalk3.default.cyan(`$${model.pricing.output.toFixed(2)} output`)} ${import_chalk3.default.dim("(per 1M tokens)")}
+`);
+    if (model.pricing.cachedInput !== void 0) {
+      stream2.write(`  ${import_chalk3.default.dim("Cached Input:")} ${import_chalk3.default.cyan(`$${model.pricing.cachedInput.toFixed(2)} per 1M tokens`)}
+`);
+    }
+    if (model.knowledgeCutoff) {
+      stream2.write(`  ${import_chalk3.default.dim("Knowledge:")}    ${model.knowledgeCutoff}
+`);
+    }
+    const features = [];
+    if (model.features.streaming) features.push("streaming");
+    if (model.features.functionCalling) features.push("function-calling");
+    if (model.features.vision) features.push("vision");
+    if (model.features.reasoning) features.push("reasoning");
+    if (model.features.structuredOutputs) features.push("structured-outputs");
+    if (model.features.fineTuning) features.push("fine-tuning");
+    if (features.length > 0) {
+      stream2.write(`  ${import_chalk3.default.dim("Features:")}     ${import_chalk3.default.blue(features.join(", "))}
+`);
+    }
+    if (model.metadata) {
+      if (model.metadata.family) {
+        stream2.write(`  ${import_chalk3.default.dim("Family:")}       ${model.metadata.family}
+`);
+      }
+      if (model.metadata.releaseDate) {
+        stream2.write(`  ${import_chalk3.default.dim("Released:")}     ${model.metadata.releaseDate}
+`);
+      }
+      if (model.metadata.notes) {
+        stream2.write(`  ${import_chalk3.default.dim("Notes:")}        ${import_chalk3.default.italic(model.metadata.notes)}
+`);
+      }
+    }
+  }
+  stream2.write("\n");
+}
+function renderJSON(models, stream2) {
+  const output = {
+    models: models.map((model) => ({
+      provider: model.provider,
+      modelId: model.modelId,
+      displayName: model.displayName,
+      contextWindow: model.contextWindow,
+      maxOutputTokens: model.maxOutputTokens,
+      pricing: {
+        input: model.pricing.input,
+        output: model.pricing.output,
+        cachedInput: model.pricing.cachedInput,
+        currency: "USD",
+        per: "1M tokens"
+      },
+      knowledgeCutoff: model.knowledgeCutoff,
+      features: model.features,
+      metadata: model.metadata
+    })),
+    shortcuts: MODEL_ALIASES
+  };
+  stream2.write(JSON.stringify(output, null, 2) + "\n");
+}
+function formatTokens(count) {
+  if (count >= 1e6) {
+    return `${(count / 1e6).toFixed(1)}M tokens`;
+  } else if (count >= 1e3) {
+    return `${(count / 1e3).toFixed(0)}K tokens`;
+  } else {
+    return `${count} tokens`;
+  }
+}
+function registerModelsCommand(program, env) {
+  program.command(COMMANDS.models).description("List all available LLM models with pricing and capabilities.").option("--provider <name>", "Filter by provider (openai, anthropic, gemini)").option("--format <format>", "Output format: table or json", "table").option("--verbose", "Show detailed model information", false).action(
+    (options) => executeAction(
+      () => handleModelsCommand(options, env),
+      env
+    )
+  );
+}
+
 // src/cli/environment.ts
 var import_node_readline = __toESM(require("readline"), 1);
-var import_chalk3 = __toESM(require("chalk"), 1);
+var import_chalk4 = __toESM(require("chalk"), 1);
 init_client();
 init_logger();
 var LOG_LEVEL_MAP = {
@@ -5327,14 +5543,14 @@ function createPromptFunction(stdin, stdout) {
         output: stdout
       });
       stdout.write("\n");
-      stdout.write(`${import_chalk3.default.cyan("\u2500".repeat(60))}
+      stdout.write(`${import_chalk4.default.cyan("\u2500".repeat(60))}
 `);
-      stdout.write(import_chalk3.default.cyan.bold("\u{1F916} Agent asks:\n"));
+      stdout.write(import_chalk4.default.cyan.bold("\u{1F916} Agent asks:\n"));
       stdout.write(`${question}
 `);
-      stdout.write(`${import_chalk3.default.cyan("\u2500".repeat(60))}
+      stdout.write(`${import_chalk4.default.cyan("\u2500".repeat(60))}
 `);
-      rl.question(import_chalk3.default.green.bold("You: "), (answer) => {
+      rl.question(import_chalk4.default.green.bold("You: "), (answer) => {
         rl.close();
         resolve(answer);
       });
@@ -5377,6 +5593,7 @@ function createProgram(env) {
   });
   registerCompleteCommand(program, env);
   registerAgentCommand(program, env);
+  registerModelsCommand(program, env);
   return program;
 }
 async function runCLI(overrides = {}) {