@noobdemon/noob-cli 1.9.8 → 1.9.10

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@noobdemon/noob-cli",
-  "version": "1.9.8",
+  "version": "1.9.10",
   "publishConfig": {
     "access": "public"
   },

package/src/agent.js

@@ -33,7 +33,7 @@ Available tools (each is self-contained; pick the SMALLEST tool that answers the
 Context is finite. Don't slurp the whole repo up front. Discover information progressively: list_dir/glob to map → grep to locate → read_file (with offset+limit for big files) to inspect only what matters. Each tool result spends your attention budget — make every call earn it. When a tool returns a huge blob, extract the few facts you need, then move on; don't re-read it later (the result stays in history).
 
 # Rules
-- TODO-BASED EXECUTION: For any multi-step task (3+ actions), CREATE a todo list FIRST in your response text (NOT in a file — just write them as "- [ ] task name" in your reply). Then WORK THROUGH EVERY ITEM, checking them off ("- [x] task name") as you complete each. The runtime parses your text for these markers and shows a progress bar to the user. BEFORE summarizing or claiming "done", verify ALL items are "- [x]". If ANY remain unchecked, CONTINUE — do not stop. NEVER stop mid-plan.
+- TODO-BASED EXECUTION: For any multi-step task (3+ actions), you MUST do TWO things in the SAME response: (1) write the todo list as "- [ ] task" markers, and (2) IMMEDIATELY start working on the FIRST unchecked item — do NOT stop after writing the list. The todo list is NOT output — it is a planning step. NEVER output a todo list and then stop. After checking off an item ("- [x]"), CONTINUE to the next unchecked item in the SAME response. Only stop when ALL items are "- [x]" or you are waiting for a tool result. Before claiming "done", verify ALL items are checked. NEVER stop mid-plan.
 - GROUND TRUTH = real TOOL RESULTs in this conversation, not your memory or what you intended to do. A file changed only if a write_file/edit_file result confirms it (see the FILES CHANGED list). A test passed / build succeeded / command worked only if a run_command result above shows it. Never narrate outcomes you didn't observe; if you haven't checked, say so and check now (read_file / list_dir / run the command). Before any "done/summary" reply, reconcile every file and result you're about to claim against the actual tool results above — if it isn't there, you didn't do it yet.
 - Investigate before editing: read the relevant files first; never invent file contents.
 - Make the smallest change that fully solves the task. Match the surrounding code style.
@@ -513,6 +513,9 @@ export async function runAgent({ history, model, signal, onTool, onStatus, onDel
 
     const system = buildSystem(history, extraToolsDoc, goal, recentSessions);
     const message = buildUserMessage(history);
+    // Cập nhật context tokens realtime — đếm system + message sau khi build xong.
+    // Mỗi iteration (tool call, steer) history thay đổi → size thay đổi → cần update.
+    tokenMeter?.setContext(countTokens(system) + countTokens(message));
     tokenMeter?.addInput(countTokens(message));
     onStatus?.("thinking");
     onDelta?.({ type: "step-start" });