feat: render LLM responses as markdown in terminal

Use rich to render markdown formatting (headings, code blocks, bold,
lists, etc.) for all LLM responses. /copy still copies the raw LLM
response text.

- Add rich>=13.0.0 dependency
- New iclaw/renderer.py with print_markdown() using rich Markdown
- Main chat, /search, and /paste all render via print_markdown()
- Streaming paths collect silently then render at end

Author: lzwjava

iclaw/main.py

@@ -22,6 +22,7 @@
 from iclaw.commands.read import handle_read_command
 from iclaw.commands.search_provider import handle_search_provider_command
 from iclaw.commands.utils import handle_copy_command
+from iclaw.renderer import print_markdown
 from iclaw.completer import IclawCompleter
 from iclaw.config import (
     CONFIG_PATH,
@@ -244,9 +245,9 @@ async def _main():
                 print()
                 reply = ""
                 for chunk in chunks:
-                    print(chunk, end="", flush=True)
                     reply += chunk
-                print("\n")
+                print_markdown(reply)
+                print()
                 messages.append({"role": "assistant", "content": reply})
                 last_reply = reply
             except UnsupportedModelError as e:
@@ -402,9 +403,9 @@ async def _main():
                 print()
                 reply = ""
                 for chunk in chunks:
-                    print(chunk, end="", flush=True)
                     reply += chunk
-                print("\n")
+                print_markdown(reply)
+                print()
                 messages.append({"role": "assistant", "content": reply})
                 last_reply = reply
             except UnsupportedModelError as e:
@@ -554,7 +555,9 @@ async def _main():
             reply = response_message["content"]
             messages.append({"role": "assistant", "content": reply})
             last_reply = reply
-            log.log_info(f"\n{reply}\n")
+            print()
+            print_markdown(reply)
+            print()
         except UnsupportedModelError as e:
             print(f"Error: {e}", file=sys.stderr)
             print("Please select a different model with /model", file=sys.stderr)

iclaw/renderer.py

@@ -0,0 +1,22 @@
+"""Markdown rendering for terminal output using rich."""
+
+from rich.console import Console
+from rich.markdown import Markdown
+
+
+def print_markdown(text: str) -> None:
+    """Render markdown text to the terminal with formatting.
+
+    Uses rich to render headings, code blocks, lists, bold/italic,
+    links, tables, etc. Falls back to plain print if rendering fails.
+    """
+    if not text or not text.strip():
+        print(text or "")
+        return
+
+    try:
+        console = Console(soft_wrap=True)
+        console.print(Markdown(text))
+    except Exception:
+        # Fallback to plain text if rich rendering fails
+        print(text)

pyproject.toml

@@ -8,7 +8,7 @@ version = "0.1.0"
 description = "Interactive CLI REPL for chatting with GitHub Copilot"
 readme = "README.md"
 requires-python = ">=3.8"
-dependencies = ["requests>=2.32.0", "pyperclip>=1.8.0", "beautifulsoup4>=4.12.0", "readability-lxml>=0.8.1", "lxml>=5.0.0", "tavily-python>=0.3.0", "prompt-toolkit>=3.0.0", "playwright>=1.40.0"]
+dependencies = ["requests>=2.32.0", "pyperclip>=1.8.0", "beautifulsoup4>=4.12.0", "readability-lxml>=0.8.1", "lxml>=5.0.0", "tavily-python>=0.3.0", "prompt-toolkit>=3.0.0", "playwright>=1.40.0", "rich>=13.0.0"]
 
 [project.scripts]
 iclaw = "iclaw.main:main"