Add service-log following and restore compatible test installs

.github/workflows/tests.yml

@@ -0,0 +1,22 @@
+name: Tests
+
+on:
+  pull_request:
+  push:
+    branches: [main]
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - uses: actions/checkout@v6
+      - uses: actions/setup-python@v6
+        with:
+          python-version: '3.12'
+      - run: python -m pip install '.[dev]' build
+      - run: PYTHONPATH=src python -m pytest
+      - run: python -m build

README.md

@@ -252,7 +252,7 @@ baton taint clear                        # remove all canary data
 ### Service Logs
 
 ```bash
-baton logs [--node N] [--level L]        # show captured service logs
+baton logs [-f | --follow] [--node N] [--level L]  # show or follow captured service logs
 ```
 
 ### Arbiter & Constrain

docs/index.html

@@ -444,7 +444,7 @@ <h2>Five minutes to a running circuit</h2>
 <span class="cmd">baton test</span> <span class="flag">--canary</span> <span class="flag">--tiers</span> <span class="arg">PII</span> <span class="flag">--duration</span> <span class="arg">5m</span>
 
 <span class="comment"># Service logs and Arbiter trust</span>
-<span class="cmd">baton logs</span> <span class="flag">--node</span> <span class="arg">api</span> <span class="flag">--level</span> <span class="arg">error</span>
+<span class="cmd">baton logs</span> <span class="flag">--follow</span> <span class="flag">--node</span> <span class="arg">api</span> <span class="flag">--level</span> <span class="arg">error</span>
 <span class="cmd">baton trust</span> <span class="arg">api</span>
 
 <span class="comment"># Build images and deploy to GCP Cloud Run</span>
@@ -764,7 +764,7 @@ <h3 style="font-size: 16px; font-weight: 600; margin: 24px 0 12px; color: var(--
 <span class="cmd">baton metrics</span> <span class="flag">--prometheus</span>                <span class="comment"># Prometheus text exposition format</span>
 <span class="cmd">baton signals</span> <span class="flag">[--node N] [--path P]</span>      <span class="comment"># recent request signals</span>
 <span class="cmd">baton signals</span> <span class="flag">--stats</span>                     <span class="comment"># per-path statistics</span>
-<span class="cmd">baton logs</span> <span class="flag">[--node N] [--level L]</span>        <span class="comment"># captured service logs</span>
+<span class="cmd">baton logs</span> <span class="flag">[-f | --follow] [--node N] [--level L]</span> <span class="comment"># show or follow captured service logs</span>
     </div>
 
     <h3 style="font-size: 16px; font-weight: 600; margin: 24px 0 12px; color: var(--accent);">Taint Analysis</h3>

pyproject.toml

@@ -25,7 +25,7 @@ dependencies = [
 dev = [
     "pytest>=8.0",
     "pytest-asyncio>=0.23",
-    "mcp[cli]>=1.26.0",
+    "mcp[cli]>=1.26.0,<2",
 ]
 gcp = [
     "google-cloud-run>=0.10",
@@ -36,7 +36,7 @@ aws = [
 ]
 otel = []
 mcp = [
-    "mcp[cli]>=1.26.0",
+    "mcp[cli]>=1.26.0,<2",
 ]
 certs = [
     "cryptography>=41.0",

src/baton/cli.py

@@ -378,6 +378,7 @@ def main(argv: list[str] | None = None) -> int:
     p_logs.add_argument("--node", default="", help="Filter by node")
     p_logs.add_argument("--level", default="", help="Minimum severity level (debug/info/warning/error/critical)")
     p_logs.add_argument("--last", type=int, default=50, help="Number of entries (default: 50)")
+    p_logs.add_argument("-f", "--follow", action="store_true", help="Follow new log entries")
     p_logs.add_argument("--dir", default=".", help="Project directory")
     p_logs.set_defaults(func=_cmd_logs)
 
@@ -1514,6 +1515,17 @@ def _cmd_dora(args: argparse.Namespace) -> int:
 def _cmd_logs(args: argparse.Namespace) -> int:
     from baton.service_log import ServiceLogCollector
 
+    if getattr(args, "follow", False):
+        records = ServiceLogCollector.follow_history(
+            args.dir,
+            node=args.node or None,
+            severity=args.level or None,
+            last_n=args.last,
+        )
+        for record in records:
+            _print_service_log(record)
+        return 0
+
     records = ServiceLogCollector.load_history(
         args.dir,
         node=args.node or None,
@@ -1524,17 +1536,22 @@ def _cmd_logs(args: argparse.Namespace) -> int:
         print("No service logs found")
         return 0
 
-    for r in records:
-        sev = r.get("severity", "info").upper()
-        node = r.get("node_name", "?")
-        ts = r.get("timestamp", "")[:19]  # Trim to seconds
-        stream = r.get("stream", "")
-        msg = r.get("message", "")
-        print(f"[{ts}] [{sev:<8}] [{node}:{stream}] {msg}")
+    for record in records:
+        _print_service_log(record)
 
     return 0
 
 
+def _print_service_log(record: dict) -> None:
+    """Print one structured service log record."""
+    sev = record.get("severity", "info").upper()
+    node = record.get("node_name", "?")
+    ts = record.get("timestamp", "")[:19]  # Trim to seconds
+    stream = record.get("stream", "")
+    msg = record.get("message", "")
+    print(f"[{ts}] [{sev:<8}] [{node}:{stream}] {msg}", flush=True)
+
+
 async def _cmd_dashboard(args: argparse.Namespace) -> int:
     import json as json_mod
     from baton.dashboard import collect, format_table

src/baton/service_log.py

@@ -7,11 +7,14 @@
 
 from __future__ import annotations
 
+import json
 import re
+import time
+from collections.abc import Iterator
 from datetime import datetime, timezone
 from pathlib import Path
 
-from baton.state import append_jsonl, read_jsonl
+from baton.state import BATON_DIR, append_jsonl, read_jsonl
 
 LOGS_FILE = "service_logs.jsonl"
 
@@ -40,6 +43,20 @@ def parse_severity(line: str) -> str:
     return ""  # caller decides default based on stream
 
 
+def _matches_filters(
+    record: dict,
+    node: str | None = None,
+    severity: str | None = None,
+) -> bool:
+    """Return whether a structured log record matches CLI filters."""
+    if node and record.get("node_name") != node:
+        return False
+    if severity and severity in SEVERITIES:
+        minimum = SEVERITIES.index(severity)
+        return SEVERITIES.index(record.get("severity", "info")) >= minimum
+    return True
+
+
 class ServiceLogCollector:
     """Captures and structures service log output.
 
@@ -113,12 +130,55 @@ def load_history(
     ) -> list[dict]:
         """Read from .baton/service_logs.jsonl."""
         records = read_jsonl(project_dir, LOGS_FILE, last_n=last_n)
-        if node:
-            records = [r for r in records if r.get("node_name") == node]
-        if severity and severity in SEVERITIES:
-            sev_idx = SEVERITIES.index(severity)
-            records = [
-                r for r in records
-                if SEVERITIES.index(r.get("severity", "info")) >= sev_idx
-            ]
-        return records
+        return [
+            record for record in records
+            if _matches_filters(record, node=node, severity=severity)
+        ]
+
+    @staticmethod
+    def follow_history(
+        project_dir: str | Path,
+        node: str | None = None,
+        severity: str | None = None,
+        last_n: int | None = 50,
+        poll_interval: float = 0.2,
+    ) -> Iterator[dict]:
+        """Yield recent records, then wait for and yield newly appended records."""
+        path = Path(project_dir) / BATON_DIR / LOGS_FILE
+        while not path.exists():
+            time.sleep(poll_interval)
+
+        with open(path) as log_file:
+            history: list[str] = []
+            while True:
+                offset = log_file.tell()
+                line = log_file.readline()
+                if not line:
+                    break
+                if not line.endswith("\n"):
+                    log_file.seek(offset)
+                    break
+                history.append(line)
+
+            if last_n is not None:
+                history = history[-last_n:]
+            for line in history:
+                if not line.strip():
+                    continue
+                record = json.loads(line)
+                if _matches_filters(record, node=node, severity=severity):
+                    yield record
+
+            while True:
+                offset = log_file.tell()
+                line = log_file.readline()
+                if not line or not line.endswith("\n"):
+                    log_file.seek(offset)
+                    time.sleep(poll_interval)
+                    continue
+
+                if not line.strip():
+                    continue
+                record = json.loads(line)
+                if _matches_filters(record, node=node, severity=severity):
+                    yield record

tests/test_cli.py

@@ -633,6 +633,57 @@ def test_signals_path_filter(self, project_dir: Path, capsys):
         assert "/users" not in out
 
 
+class TestLogsCLI:
+    @pytest.mark.parametrize("follow_flag", ["-f", "--follow"])
+    def test_follow_streams_records_with_filters(
+        self,
+        project_dir: Path,
+        monkeypatch,
+        capsys,
+        follow_flag: str,
+    ):
+        from baton.service_log import ServiceLogCollector
+
+        captured = {}
+
+        def follow_history(project_dir, *, node, severity, last_n):
+            captured.update(
+                project_dir=project_dir,
+                node=node,
+                severity=severity,
+                last_n=last_n,
+            )
+            return iter([
+                {
+                    "node_name": "api",
+                    "stream": "stderr",
+                    "severity": "error",
+                    "message": "request failed",
+                    "timestamp": "2026-01-01T00:00:00+00:00",
+                }
+            ])
+
+        monkeypatch.setattr(ServiceLogCollector, "follow_history", follow_history)
+
+        rc = main([
+            "logs",
+            follow_flag,
+            "--node", "api",
+            "--level", "warning",
+            "--last", "7",
+            "--dir", str(project_dir),
+        ])
+
+        assert rc == 0
+        assert captured == {
+            "project_dir": str(project_dir),
+            "node": "api",
+            "severity": "warning",
+            "last_n": 7,
+        }
+        assert "[ERROR   ] [api:stderr] request failed" in capsys.readouterr().out
+
+
 class TestMetricsCLI:
     def _init_with_metrics(self, d: Path) -> Path:
         """Initialize a project and write telemetry data."""