A self-improving RLM agent for coding workflows and long-running autonomous tasks.
Rust
simplify daily stats Slack formatting (#3503)
kt1a4dde57b198
diff --git a/docs/daily-stats.md b/docs/daily-stats.mdindex 366ff7b0ca..c4f084f217 100644--- a/docs/daily-stats.md+++ b/docs/daily-stats.md@@ -11,6 +11,10 @@ the preceding seven-day window, and MAU with the preceding 30-day window. Succesrate changes use percentage points. A zero baseline shows `new (baseline 0)`;no completed runs gives an unavailable success rate.+The Slack message marks increases with π’ β and decreases with π΄ β; unchanged+values keep a neutral arrow. These indicate direction, not whether a change is+good or bad. A divider separates the metrics from the dashboard link.+Active users count anonymous installations that emitted `agent run completed`,including successful, failed, and aborted runs. They do not count verified people.Tokens sum `total_tokens`, including recorded cache usage; they do not measurediff --git a/scripts/daily_stats.py b/scripts/daily_stats.pyindex f389bd29e4..b875fc41e7 100644--- a/scripts/daily_stats.py+++ b/scripts/daily_stats.py@@ -115,7 +115,7 @@ def change(current, previous, *, percentage_points=False):else:delta, unit = (current / previous - 1) * 100, "%"delta = round(delta, 1)- arrow = "β" if delta > 0 else "β" if delta < 0 else "β"+ arrow = "π’ β" if delta > 0 else "π΄ β" if delta < 0 else "β"return f"{arrow} {abs(delta):.1f}{unit}"@@ -148,10 +148,7 @@ def render_report(points, day):lines.append(f"β’ *Run success rate: {rate}* Β· "f"{change(rates[day], rates[yesterday], percentage_points=True)} vs yesterday Β· "f"{change(rates[day], rates[last_week], percentage_points=True)} vs last week")- lines.extend([f"<{DASHBOARD_URL}|Open dashboard>",- "Active users count anonymous installations with a completed run. "- "Tokens are recorded usage, including cache; they do not measure cost. "- "Test accounts and known benchmarks are excluded."])+ lines.extend(["", f"<{DASHBOARD_URL}|Open dashboard>"])return "\n".join(lines)@@ -205,7 +202,13 @@ def main():"refresh": "force_blocking", "async": False,}, token=token, service="PostHog"))text = render_report(parse_results(response, day), day)- args.output.write_text(json.dumps({"text": text, "unfurl_links": False,+ metrics, dashboard = text.rsplit("\n\n", 1)+ blocks = [+ {"type": "section", "text": {"type": "mrkdwn", "text": metrics}},+ {"type": "divider"},+ {"type": "section", "text": {"type": "mrkdwn", "text": dashboard}},+ ]+ args.output.write_text(json.dumps({"text": text, "blocks": blocks, "unfurl_links": False,"unfurl_media": False}))print("Report prepared and validated. Metric values were not logged.")elif args.command == "send":diff --git a/scripts/test_daily_stats.py b/scripts/test_daily_stats.pyindex 037f1555ba..fb4b0b9b1d 100644--- a/scripts/test_daily_stats.py+++ b/scripts/test_daily_stats.py@@ -119,20 +119,21 @@ class ReportTests(unittest.TestCase):stats.parse_results(response, DAY)def test_percentage_changes_and_zero_baselines(self):- cases = [(100, 80, "β 25.0%"), (80, 100, "β 20.0%"),+ cases = [(100, 80, "π’ β 25.0%"), (80, 100, "π΄ β 20.0%"),(100, 100, "β 0.0%"), (0, 0, "β 0.0%"),(100, 0, "new (baseline 0)"), (None, 100, "n/a")]for current, previous, expected in cases:self.assertEqual(stats.change(current, previous), expected)- self.assertEqual(stats.change(90, 80, percentage_points=True), "β 10.0pp")+ self.assertEqual(stats.change(90, 80, percentage_points=True), "π’ β 10.0pp")+ self.assertEqual(stats.change(80, 90, percentage_points=True), "π΄ β 10.0pp")def test_report_compares_complete_rolling_windows_and_success_rate(self):report = stats.render_report(stats.parse_results(fixture(), DAY), DAY)self.assertTrue(report.startswith("*Prime Agent Daily Stats (Oct 5, 2026)*\n"))lines = report.splitlines()- self.assertTrue(lines[2].startswith("β’ *DAU: 100* Β· β 25.0%"))- self.assertTrue(lines[3].startswith("β’ *WAU: 300* Β· β 50.0%"))- self.assertTrue(lines[4].startswith("β’ *MAU: 700* Β· β 40.0%"))+ self.assertTrue(lines[2].startswith("β’ *DAU: 100* Β· π’ β 25.0%"))+ self.assertTrue(lines[3].startswith("β’ *WAU: 300* Β· π’ β 50.0%"))+ self.assertTrue(lines[4].startswith("β’ *MAU: 700* Β· π’ β 40.0%"))self.assertNotIn("β", report)self.assertNotIn("(UTC)", report)for expected in ("DAU: 100", "β 25.0% vs yesterday", "β 100.0% vs last week",@@ -163,7 +164,20 @@ class ReportTests(unittest.TestCase):request.assert_not_called()send.assert_not_called()self.assertNotIn("95.0%", stdout.getvalue())- self.assertIn("Run success rate: 95.0%", json.loads(output.read_text())["text"])+ payload = json.loads(output.read_text())+ text = stats.render_report(stats.parse_results(fixture(), DAY), DAY)+ metrics, dashboard = text.rsplit("\n\n", 1)+ self.assertEqual(payload, {+ "text": text, "unfurl_links": False, "unfurl_media": False,+ "blocks": [+ {"type": "section", "text": {"type": "mrkdwn", "text": metrics}},+ {"type": "divider"},+ {"type": "section", "text": {"type": "mrkdwn", "text": dashboard}},+ ],+ })+ self.assertIn("Run success rate: 95.0%", payload["text"])+ self.assertTrue(payload["text"].endswith(f"<{stats.DASHBOARD_URL}|Open dashboard>"))+ self.assertNotIn("Active users count", payload["text"])def test_prepare_api_uses_fresh_blocking_query(self):with tempfile.TemporaryDirectory() as directory, \