agent.libx.js 0.93.36 → 0.93.38

package/README.md

@@ -117,7 +117,7 @@ agentx --resume <id> "…"    # resume a specific session
 - **Any provider** — set `ANTHROPIC_API_KEY` / `OPENAI_API_KEY` / `GOOGLE_API_KEY` / `GROQ_API_KEY`; choose with `-m provider/model`.
 - **@-file mentions & headless JSON** — reference files inline in a prompt with `@path` (e.g. `explain @src/Agent.ts`); script with `-p --output-format json` to get one machine-readable result object on stdout (activity stays on stderr).
 - **Tab-completion** — `Tab` completes `/<command>` names and `@<path>` file/dir references (descends subdirs, dotfiles hidden unless typed) straight from the working tree.
-- **Duplex mode** — `agentx --duplex` runs the full standard REPL (slash commands, sessions, postures, rewind, MCP) with the dual-model engine driving turns: a fast voice model (`--voice-model`, default haiku) answers every line instantly and delegates real work to background workers built with the same wiring as a normal run (fs mode, permissions, MCP); worker activity shows as dim chrome and results are re-voiced when ready.
+- **Duplex mode** — `agentx --duplex` runs the full standard REPL (slash commands, sessions, postures, rewind, MCP) with the three-tier engine driving turns: a fast voice model (`--voice-model`, default `groq/openai/gpt-oss-120b`) answers every line instantly and delegates real work to background workers built with the same wiring as a normal run (fs mode, permissions, MCP); worker activity shows as dim chrome and results are re-voiced when ready. Switch any tier live with `/model` (opens a reflex/act/think picker), or the `/voice-model` · `/think-model` shortcuts.
 - **MCP servers** — declare `mcpServers: { name: { command, args } | { url } }` in config and they're auto-mounted at startup (in parallel, with an optional `mountTimeoutMs` deadline so one slow/dead server never blocks the rest): the client does the JSON-RPC handshake (stdio or HTTP) + `tools/list`, and the discovered tools appear as `mcp__<name>__<tool>` in `/tools` (inspect with `/mcp`). A bad server is logged and skipped, never blocking the agent. For large tool sets, **deferred mode** (`makeMcpToolSearch` / `mountMcpDeferred`) exposes just two bounded tools (`ToolSearch` + `McpCall`) instead of N defs — dodging the provider tool-cap and improving selection accuracy. **`mountMcpCatalog`** goes further: a cached, hash-keyed catalog + lazy connect means a turn that uses no MCP tool opens **zero** connections, and one that uses a tool connects exactly that server — latency scales with tools-used, not servers-configured. A down server is **negative-cached** (`failureCooldownMs`) so it never re-floors a later turn at the deadline. For zero turn-path latency even on a cold process, call **`warmMcpCatalog`** at boot + on a timer (off-turn discovery) and mount with **`{ discover: 'cache-only' }`** — the turn then never synchronously connects: it serves the warmed catalog and discovers any miss in the background.
 
 ## 🧬 It improves itself

package/dist/cli.js

@@ -9104,6 +9104,47 @@ async function repl(args, ai, cfg, cwd) {
     persistModel(cwd, m);
     err(dim("  model \u2192 " + m + "\n"));
   };
+  const applyReflexModel = (id) => {
+    const m = resolveModelOrNewest(id);
+    dx.options.reflexModel = m;
+    dx.voice.options.model = m;
+    err(green(`  \u2713 reflex model \u2192 ${m}
+`));
+  };
+  const applyThinkModel = (id) => {
+    const m = resolveModelOrNewest(id);
+    dx.setThinkModel(m);
+    err(green(`  \u2713 think model \u2192 ${m}
+`));
+  };
+  const pickModelByTier = async () => {
+    const thinkLabel = dx.options.thinkModel === false ? "off" : String(dx.options.thinkModel);
+    const tier = await selectMenu(process.stderr, { title: "Set model for which tier?", items: [
+      { label: "Reflex (voice)", value: "reflex", desc: dx.options.reflexModel },
+      { label: "Act (worker)", value: "act", desc: work.model },
+      { label: "Think", value: "think", desc: thinkLabel }
+    ] });
+    if (!tier) return;
+    if (tier === "think") {
+      const sub = await selectMenu(process.stderr, { title: "Think tier", items: [
+        { label: "Choose a model\u2026", value: "__pick__", desc: thinkLabel === "off" ? "currently off" : `current: ${thinkLabel}` },
+        { label: "Disable Think tier", value: "__off__" }
+      ] });
+      if (!sub) return;
+      if (sub === "__off__") {
+        dx.setThinkModel(false);
+        err(green("  \u2713 think tier disabled\n"));
+        return;
+      }
+      const picked2 = await pickModel(dx.options.thinkModel === false ? "anthropic/claude-opus-4-6" : String(dx.options.thinkModel));
+      if (picked2) applyThinkModel(picked2);
+      return;
+    }
+    const picked = await pickModel(tier === "reflex" ? dx.options.reflexModel : work.model);
+    if (!picked) return;
+    if (tier === "reflex") applyReflexModel(picked);
+    else setModel(picked);
+  };
   const addWorkTools = (ts) => {
     if (duplex) work.tools = [...work.tools ?? [], ...ts];
     else agent.addTools(ts);
@@ -9480,15 +9521,19 @@ ${extra}` : body);
       }
     },
     model: {
-      desc: "switch model \u2014 /model <id>, or /model alone for an interactive picker (duplex: the worker model)",
+      desc: "switch model \u2014 /model <id> (duplex: the worker), or /model alone for a picker (duplex: pick a tier first)",
       run: async (a) => {
         if (a[0]) {
           setModel(a[0]);
           return;
         }
+        if (duplex) {
+          await pickModelByTier();
+          return;
+        }
         const picked = await pickModel(work.model);
         if (picked) setModel(picked);
-        else err(dim("  " + (duplex ? `reflex ${dx.options.reflexModel} \xB7 act ${work.model}${dx.options.thinkModel !== false ? ` \xB7 think ${dx.options.thinkModel}` : ""}` : work.model) + "\n"));
+        else err(dim("  " + work.model + "\n"));
       }
     },
     ...duplex ? { workers: {
@@ -9515,19 +9560,12 @@ ${extra}` : body);
     }, "voice-model": {
       desc: "switch the reflex (voice) model \u2014 /voice-model <id>, or alone for a picker",
       run: async (a) => {
-        const apply = (id) => {
-          const m = resolveModelOrNewest(id);
-          dx.options.reflexModel = m;
-          dx.voice.options.model = m;
-          err(green(`  \u2713 reflex model \u2192 ${m}
-`));
-        };
         if (a[0]) {
-          apply(a[0]);
+          applyReflexModel(a[0]);
           return;
         }
         const picked = await pickModel(dx.options.reflexModel);
-        if (picked) apply(picked);
+        if (picked) applyReflexModel(picked);
         else err(dim(`  reflex ${dx.options.reflexModel}
 `));
       }
@@ -9540,19 +9578,13 @@ ${extra}` : body);
 `));
           return;
         }
-        const apply = (id) => {
-          const m = resolveModelOrNewest(id);
-          dx.setThinkModel(m);
-          err(green(`  \u2713 think model \u2192 ${m}
-`));
-        };
         if (a[0]) {
-          apply(a[0]);
+          applyThinkModel(a[0]);
           return;
         }
         const current = dx.options.thinkModel === false ? void 0 : dx.options.thinkModel;
         const picked = await pickModel(current ?? "anthropic/claude-opus-4-6");
-        if (picked) apply(picked);
+        if (picked) applyThinkModel(picked);
         else err(dim(`  think ${dx.options.thinkModel === false ? "off" : dx.options.thinkModel}
 `));
       }