regcode 0.8.0__tar.gz → 0.9.0__tar.gz

{regcode-0.8.0 → regcode-0.9.0}/.gitignore

@@ -11,4 +11,6 @@ wheels/
 config.yaml
 review.py
 *.config.yaml
-.env
+.env
+debug*.*
+llm_calls*.json

{regcode-0.8.0 → regcode-0.9.0}/GOALS.md

@@ -530,10 +530,15 @@ All phases complete. 50/50 tests passing. Lint clean.
 
 ## Completed (2026-06-30 — TUI rewrite)
 
-Complete rewrite of `tui.py` from scratch using Textual's `RichLog` widget instead of `TextArea`:
+## Completed (2026-07-05)
 
-1. ✅ **`RichLog` instead of `TextArea`** — RichLog natively renders Rich markup (`[bold]`, `[dim]`, etc.) so no fragile Rich→ANSI conversion is needed. Users see properly colored, styled text. RichLog also supports scrollback selection/copy via terminal features.
-2. ✅ **Clean Catppuccin Mocha dark theme** — uses the Catppuccin color palette (`#1e1e2e` background, `#cdd6f4` text, `#313244` surfaces, `#89b4fa` accents) for a modern, polished look.
-3. ✅ **Proper three-zone layout** — status bar (version | model | cost | tokens), scrollable chat history, bottom input bar with progress indicator. No raw markup tags visible.
-4. ✅ **Same backward-compatible API** — `ChatUI` facade, `ChatApp` class, `_ColorFormatter`, all callback methods (`handle_status`, `handle_result`, `handle_cost`, `print_text_chunk`, etc.) preserved exactly as before for CLI integration.
-5. ✅ **All 164 tests pass**, lint clean.
+1. ✅ **Agent loop rewrite** — `Agent.chat()` completely rewritten following `debug.ipynb` pattern. Key changes:
+   - Extracted `_chat_loop()` for non-streaming mode, `_stream()` for streaming — both use shared `_execute_tool_calls()` helper (eliminating ~300 lines of duplicated tool execution code).
+   - Added `on_message` callback parameter: called after every LLM call with the assistant's text content BEFORE tool results are shown. This fixes the broken chain where "Let me search" was never displayed — the user now sees it in real-time as tools execute.
+   - Fixed `_stream()`: removed `full_text = ""` reset at each iteration (was wiping previously accumulated text). Now `full_text` accumulates across ALL iterations so the user sees the complete conversation.
+   - Fixed `_chat_loop` loop bound: was using `self.config.tool_budget` (default 20) instead of `max_tool_calls` (default 10), causing excessive LLM calls. Now properly bounded by `max_tool_calls`.
+   - When `finish_reason == "stop"` or text + no tools, loop breaks immediately — no more dead-end loop iterations.
+
+2. ✅ **TUI integration** — `_run_agent_sync()` and `_run_agent_multimodal()` now pass `on_message` callback to `agent.chat()`. The callback inserts each assistant text response directly into the chat area via `call_from_thread(self._insert_rich, text)`, so the user sees intermediate thinking ("Let me search the internet...") BEFORE tool execution markers appear. This makes the agent feel alive and responsive instead of silent during multi-tool chains.
+
+3. ✅ **All 193 tests pass, lint clean.**

{regcode-0.8.0 → regcode-0.9.0}/PKG-INFO

@@ -1,8 +1,9 @@
 Metadata-Version: 2.4
 Name: regcode
-Version: 0.8.0
+Version: 0.9.0
 Summary: Minimalistic coding agent with Python API and CLI
 Requires-Python: >=3.10
+Requires-Dist: boto3>=1.43.40
 Requires-Dist: click>=8.0
 Requires-Dist: crawl4ai>=0.4.0
 Requires-Dist: litellm>=1.83.17

{regcode-0.8.0 → regcode-0.9.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "regcode"
-version = "0.8.0"
+version = "0.9.0"
 description = "Minimalistic coding agent with Python API and CLI"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -17,6 +17,7 @@ dependencies = [
     "textual>=2.0.0",
     "pathspec>=1.1.1",
     "pyperclip>=1.9.0",
+    "boto3>=1.43.40",
 ]
 
 [project.scripts]

{regcode-0.8.0 → regcode-0.9.0}/regcode/__init__.py

@@ -2,4 +2,4 @@ from regcode.config import Config
 from regcode.main import Agent, AgentStatus, ToolCall
 
 __all__ = ["Agent", "Config", "AgentStatus", "ToolCall"]
-__version__ = "0.8.0"
+__version__ = "0.9.0"