feat(cli): render markdown in responses with auto-detection

[pedronalis]

Summary

Render AI responses as Rich Markdown instead of raw text, so **bold**, - lists, code blocks display properly formatted in the terminal.

Changes

cli.py

• Add _rich_markdown_from_text() using rich.markdown.Markdown for proper rendering
• Add _smart_render() that auto-detects markdown syntax and routes to appropriate renderer
• Updated 3 response rendering points (main responses, /plan, /btw)
• Falls back to _rich_text_from_ansi() for plain text (no markdown syntax detected)

tests/cli/test_markdown_rendering.py (new)

• 5 static tests verifying functions exist and are used
• 4 integration tests verifying Rich Markdown renders correctly

Approach

Unlike PR #2650, this implementation:

• No config flag needed — markdown detection is automatic
• Simpler code — 25 lines vs 199 lines
• No streaming re-render complexity — works for non-streaming responses
• Graceful fallback — plain ANSI for responses without markdown

Test plan

• Static tests pass (function existence, import checks)
• Integration tests verify Rich Markdown rendering
• Manual: run hermes and confirm markdown renders properly

Comparison to PR #2650

Aspect	#2650	This PR
Files changed	2	2
Lines	+199 / -18	+25 / -3
Commits	6	2
Config flag	No	No
Streaming support	Yes (complex)	No (simpler)
Tests	8	9

This PR is a minimal, focused implementation. Streaming support can be added later if needed.

[blasai1739217-cmyk]

Friendly bump on this PR in case it fell through the cracks \u2014 would love a review when someone has a minute. Thanks!

[teknium1]

Thanks for the contribution! This PR's core feature — rendering AI responses with rich.markdown.Markdown — is already implemented on main and shipped in v2026.4.23.

This is an automated hermes-sweeper review.

Evidence:

• cli.py:1239 — _render_final_assistant_content() uses rich.markdown.Markdown and supports three modes: render, strip, and raw
• hermes_cli/config.py:634 — display.final_response_markdown config key defaults to strip; set it to render to get full Rich Markdown output
• Commit 177e6eb3d wired markdown rendering into all three response rendering points (main responses, /btw, background tasks) including streaming
• tests/cli/test_cli_markdown_rendering.py already covers the same rendering helpers
• Shipped in release v2026.4.23

The existing implementation is a superset of this PR: it adds streaming support and a config-controlled mode rather than auto-detection heuristics. Closing as already implemented.

[alt-glitch]

Related to #12836 and #5084 — earlier Rich Markdown rendering PRs. This is a simpler alternative without streaming support.