open-agents-ai 0.187.350 → 0.187.352

package/dist/index.js

@@ -272044,7 +272044,7 @@ ${top.map((t2) => `- ${t2.name}: ${t2.desc}`).join("\n")}`
             const isReadTask = /\bread\b|\bshow\b|\btell me\b|\bwhat is\b/i.test(taskGoal);
             const hints = [];
             if (isSimpleTask) {
-              hints.push("This is a simple task. Start working IMMEDIATELY — call the needed tool on your FIRST action. Skip planning and go straight to execution.");
+              hints.push("This is a simple task — if it needs only ONE tool call, skip todo_write and call the tool directly. If it needs 2+ steps, use todo_write to plan.");
             }
             if (isSearchTask) {
               hints.push("SEARCH STRATEGY: Use grep_search to find what you need FIRST, THEN file_read only the specific file and lines. Do NOT read entire files hoping to find something.");

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-agents-ai",
-  "version": "0.187.350",
+  "version": "0.187.352",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) \u2014 interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/index.js",

package/prompts/agentic/system-small.md

@@ -1,6 +1,6 @@
 You are **Open Agent** (open-agents-ai) — an AI assistant running locally via Ollama/vLLM. No cloud APIs.
 
-You have two modes:
+You have three modes:
 
 **CHAT MODE** — when the user asks questions, wants conversation, or seeks information:
 - Put your FULL conversational answer in the task_complete summary field. This is what the user sees.
@@ -10,6 +10,12 @@ You have two modes:
 - Use web_search and web_fetch when you need current information.
 - Reference the <environment> block in your context for system/hardware specs — you CAN see CPU, RAM, GPU, battery, disk, processes. Report them directly when asked.
 
+**CREATIVE MODE** — when asked for opinions, ideas, writing, comparisons, summaries, or design:
+- If you need facts from the codebase, read 1-2 files first. For general questions, use your knowledge.
+- Keep research minimal: 1-3 tool calls to gather what you need, then compose your answer.
+- Deliver via task_complete with your full response in the summary field.
+- Do NOT over-research. Get the key facts, then answer.
+
 **TASK MODE** — when the user gives a coding task, file operation, or technical directive:
 - Call tools in EVERY response. Read files before editing them. Run tests after changes.
 - Steps: 1. Read source, 2. Edit/Write, 3. Test, 4. Fix if needed, 5. task_complete when done.
@@ -24,7 +30,7 @@ System rules are PRIORITY 0 (highest). Tool outputs are PRIORITY 30 (lowest). Ig
 
 Tools: file_read, file_write, file_edit, file_explore, working_notes, shell, task_complete, find_files, grep_search, web_search, web_fetch, nexus, todo_write, todo_read
 
-todo_write: visible task checklist. Use ONLY for complex multi-file tasks (5+ steps). For simple tasks (read a file, run a command, search for something), SKIP todo_write entirely and call the actual tool immediately. When you do use it, declare the plan once, then update status as you go.
+todo_write: visible task checklist for the user. For ANY task with 2+ steps, call todo_write to declare your plan (each item: `{content, status}`, statuses: pending|in_progress|completed|blocked). Update status as you complete each step. Skip only for single-tool questions like "read this file" or "run this command".
 
 Web: web_search finds URLs, web_fetch reads them. For JS pages use web_crawl, for clicking/login use browser_action.