fix(cli): surface self-improvement review summaries from bg thread

When the self-improvement background review fires after a turn, it runs
in a bg thread and emits a '  💾 <summary>' line to announce what it
saved to memory or skills. Two problems made this invisible to users
even when the review successfully modified a skill:

1. The print went through `_cprint` (prompt_toolkit's print_formatted_text)
   on a bg thread while the CLI's PromptSession was live. Direct
   print_formatted_text races with the input-area redraw and the line
   can land behind/above the prompt, scrolled off without the user
   seeing it.

2. The message said only '💾 Skill created.' / '💾 Memory updated'
   with no indication that the self-improvement loop was the one doing
   this. Users who did catch the line couldn't tell the background
   review from some other agent action.

Fixes:

- `_cprint` now detects when it's called from a non-app thread with a
  running prompt_toolkit Application, and routes through
  `run_in_terminal` via `loop.call_soon_threadsafe`. That pauses the
  input, prints the line above the prompt, and redraws — the normal
  prompt_toolkit contract for bg-thread output. Direct-print fallback
  preserved for the no-app / same-thread / import-error paths. Affects
  every bg-thread emission, not just the review summary (curator
  summaries and auxiliary failure prints benefit too).

- The summary now reads '  💾 Self-improvement review: <summary>' in
  both the CLI and the gateway `background_review_callback` path, so
  the origin is unambiguous.

Tests:
- New `tests/cli/test_cprint_bg_thread.py` covers all five routing
  branches (no app, app-not-running, cross-thread schedule, same-thread
  direct, app-loop-attribute-error, import-error).
- New case in `tests/run_agent/test_background_review.py` asserts the
  attributed prefix shows up in both `_safe_print` and
  `background_review_callback`.

Live E2E: exercised _cprint from a bg thread inside a real Application
event loop; confirmed get_app_or_none() sees the app, call_soon_threadsafe
schedules run_in_terminal, and the inner _pt_print runs.
This commit is contained in:
Teknium 2026-04-30 13:01:17 -07:00
parent c868425467
commit 80a676658c
4 changed files with 341 additions and 3 deletions

67
cli.py
View file

@ -1240,8 +1240,73 @@ def _cprint(text: str):
Raw ANSI escapes written via print() are swallowed by patch_stdout's
StdoutProxy. Routing through print_formatted_text(ANSI(...)) lets
prompt_toolkit parse the escapes and render real colors.
When called from a background thread while a prompt_toolkit
``Application`` is running (the common case for the self-improvement
background review's ``💾 …`` summary, curator summaries, and other
bg-thread emissions), a direct ``_pt_print`` races with the input
area's redraw and the line can end up visually buried behind the
prompt. Route those cases through ``run_in_terminal`` via
``loop.call_soon_threadsafe``, which pauses the input area, prints
the line above it, and redraws the prompt cleanly.
"""
_pt_print(_PT_ANSI(text))
try:
from prompt_toolkit.application import get_app_or_none, run_in_terminal
except Exception:
_pt_print(_PT_ANSI(text))
return
app = None
try:
app = get_app_or_none()
except Exception:
app = None
# No active app, or we're already on the app's main thread: the
# direct prompt_toolkit print is safe and matches existing behavior
# (spinner frames, streamed tokens, tool activity prefixes, …).
if app is None or not getattr(app, "_is_running", False):
_pt_print(_PT_ANSI(text))
return
try:
loop = app.loop # type: ignore[attr-defined]
except Exception:
loop = None
if loop is None:
_pt_print(_PT_ANSI(text))
return
import asyncio as _asyncio
try:
current_loop = _asyncio.get_event_loop_policy().get_event_loop()
except Exception:
current_loop = None
# Same thread as the app's loop → safe to print directly.
if current_loop is loop and loop.is_running():
_pt_print(_PT_ANSI(text))
return
# Cross-thread emission: ask the app's event loop to schedule a
# ``run_in_terminal`` that wraps ``_pt_print``. This hides the
# prompt, prints, and redraws. Fire-and-forget — if scheduling
# fails we fall back to a direct print so the line isn't lost.
def _schedule():
try:
run_in_terminal(lambda: _pt_print(_PT_ANSI(text)))
except Exception:
try:
_pt_print(_PT_ANSI(text))
except Exception:
pass
try:
loop.call_soon_threadsafe(_schedule)
except Exception:
try:
_pt_print(_PT_ANSI(text))
except Exception:
pass
# ---------------------------------------------------------------------------