agent-learning 0.8.0__tar.gz → 0.8.2__tar.gz

{agent_learning-0.8.0/src/agent_learning.egg-info → agent_learning-0.8.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-learning
-Version: 0.8.0
+Version: 0.8.2
 Summary: Evidence-driven reasoned and learned decisions for AI agents with inspectable autonomy.
 Author: Chris Tava
 License: MIT License

{agent_learning-0.8.0 → agent_learning-0.8.2}/README.md

@@ -82,8 +82,7 @@ points, risk floors, tier, and resolved proportional criteria.
 
 ### Windows CLI
 
-For a Python-independent installation, download `agent-learn.exe` or the
-standalone installer from the
+For a Python-independent installation, download the Windows ZIP or installer from the
 [latest GitHub release](https://github.com/microsoft/agent-learning/releases/latest).
 The installer can add its installation directory to your user `PATH`, so
 `agent-learn` works from PowerShell or Command Prompt without Python or `pip`.
@@ -101,7 +100,7 @@ or install it with the automation script:
 
 ```bash
 curl -fsSL https://raw.githubusercontent.com/microsoft/agent-learning/main/scripts/install-linux.sh -o /tmp/install-linux.sh
-bash /tmp/install-linux.sh --version 0.8.0 --install-dir /usr/local/bin
+bash /tmp/install-linux.sh --version 0.8.2 --install-dir /usr/local/bin
 ```
 
 The Linux installation guide covers Debian/Ubuntu, RHEL-compatible, and

{agent_learning-0.8.0 → agent_learning-0.8.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "agent-learning"
-version = "0.8.0"
+version = "0.8.2"
 description = "Evidence-driven reasoned and learned decisions for AI agents with inspectable autonomy."
 readme = "PYPI.md"
 requires-python = ">=3.10"

{agent_learning-0.8.0 → agent_learning-0.8.2}/src/agent_learning/_version.py

@@ -1,3 +1,3 @@
 """Package version."""
 
-__version__ = "0.8.0"
+__version__ = "0.8.2"

{agent_learning-0.8.0 → agent_learning-0.8.2}/src/agent_learning/cli.py

@@ -93,7 +93,7 @@ def _build_arg_parser() -> argparse.ArgumentParser:
         description=f"Evidence-driven decision CLI for AI agents. SDK version {__version__}.",
     )
     parser.add_argument("--version", action="version", version=f"%(prog)s {__version__}")
-    sub = parser.add_subparsers(dest="command", required=True)
+    sub = parser.add_subparsers(dest="command")
 
     sub.add_parser("list", help="List discovered agent ids and names.")
 
@@ -996,6 +996,9 @@ def main(argv: list[str] | None = None) -> int:
     logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(name)s - %(message)s")
     parser = _build_arg_parser()
     args = parser.parse_args(argv)
+    if args.command is None:
+        parser.print_help()
+        return 0
     dispatch = {
         "list": _cmd_agents_list,
         "tasks-list": _cmd_agent_tasks_list,

{agent_learning-0.8.0 → agent_learning-0.8.2}/src/agent_learning/learners/reinforce.py

@@ -53,6 +53,8 @@ class ReinforceLearner(Learner):
         for r in rewards:
             if r.source != RewardSource.AGGREGATE:
                 continue
+            if not math.isfinite(r.value) or not -1.0 <= r.value <= 1.0:
+                raise ValueError("aggregate reward value must be finite and within [-1, 1]")
             # Rescoring may append a replacement aggregate. Keep the newest.
             current = aggregate_rewards.get(r.episode_id)
             if current is None or r.created_at > current.created_at:

{agent_learning-0.8.0 → agent_learning-0.8.2/src/agent_learning.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-learning
-Version: 0.8.0
+Version: 0.8.2
 Summary: Evidence-driven reasoned and learned decisions for AI agents with inspectable autonomy.
 Author: Chris Tava
 License: MIT License

{agent_learning-0.8.0 → agent_learning-0.8.2}/tests/test_cli.py

@@ -37,6 +37,8 @@ def _full_episode() -> Episode:
 
 def test_help_and_version_print_sdk_version(capsys) -> None:
     assert f"SDK version {__version__}" in cli._build_arg_parser().format_help()
+    assert cli.main([]) == 0
+    assert "usage: agent-learn" in capsys.readouterr().out
     with pytest.raises(SystemExit) as exit_info:
         cli.main(["--version"])
     assert exit_info.value.code == 0

{agent_learning-0.8.0 → agent_learning-0.8.2}/tests/test_learner.py

@@ -12,6 +12,8 @@ from __future__ import annotations
 
 import random
 
+import pytest
+
 from agent_learning.config import LearnerConfig
 from agent_learning.learners import ReinforceLearner
 from agent_learning.policy import SoftmaxPolicy
@@ -116,3 +118,22 @@ def test_most_recent_aggregate_reward_wins() -> None:
 
     assert result.mean_reward == 0.8
     assert policy.snapshot().logits["a"] > 0.0
+
+
+@pytest.mark.parametrize("value", [float("nan"), float("inf"), -float("inf"), 1.01, -1.01])
+def test_invalid_aggregate_reward_is_rejected_without_policy_update(value: float) -> None:
+    policy = SoftmaxPolicy.from_actions([Action(id="a"), Action(id="b")])
+    learner = ReinforceLearner(LearnerConfig(learning_rate=0.5, entropy_bonus=0.0))
+    episode = _make_episode("default", "a")
+    reward = Reward(
+        episode_id=episode.id,
+        agent_id=episode.agent_id,
+        source=RewardSource.AGGREGATE,
+        value=value,
+    )
+    before = policy.snapshot()
+
+    with pytest.raises(ValueError, match=r"finite and within \[-1, 1\]"):
+        learner.update(policy, [episode], [reward])
+
+    assert policy.snapshot() == before

agent_learning-0.8.2/tests/test_release_workflow.py

@@ -0,0 +1,47 @@
+"""Release workflow contract tests."""
+
+from pathlib import Path
+
+
+def test_release_job_uses_flattened_linux_artifact_path() -> None:
+    workflow = (
+        Path(__file__).resolve().parents[1] / ".github" / "workflows" / "publish.yaml"
+    ).read_text(encoding="utf-8")
+
+    expected = 'release-assets/agent-learning-cli-${VERSION}-linux-x64.tar.gz'
+    obsolete = (
+        'release-assets/dist-installer/'
+        'agent-learning-cli-${VERSION}-linux-x64.tar.gz'
+    )
+    assert expected in workflow
+    assert obsolete not in workflow
+
+
+def test_release_job_uses_flattened_windows_artifact_paths() -> None:
+    workflow = (
+        Path(__file__).resolve().parents[1] / ".github" / "workflows" / "publish.yaml"
+    ).read_text(encoding="utf-8")
+
+    for suffix in ("exe", "zip"):
+        filename = f"agent-learning-cli-${{VERSION}}-windows-x64.{suffix}"
+        assert f'release-assets/{filename}' in workflow
+        assert f'release-assets/dist-installer/{filename}' not in workflow
+
+
+def test_windows_installer_uses_directory_bundle_and_precedes_old_shims() -> None:
+    root = Path(__file__).resolve().parents[1]
+    build_script = (root / "scripts" / "build_windows_installer.ps1").read_text(
+        encoding="utf-8"
+    )
+    installer = (
+        root / "packaging" / "windows" / "agent-learning-installer.iss"
+    ).read_text(encoding="utf-8")
+
+    assert "--onedir" in build_script
+    assert "--onefile" not in build_script
+    assert "$env:LOCALAPPDATA" in build_script
+    assert 'Source: "dist\\agent-learn\\*"' in installer
+    assert "recursesubdirs" in installer
+    assert "PrivilegesRequired=lowest" in installer
+    assert 'ValueData: "{code:PathWithEntryFirst|{app}}"' in installer
+    assert 'ValueData: "{olddata};{app}"' not in installer

agent_learning-0.8.0/tests/test_release_workflow.py

@@ -1,17 +0,0 @@
-"""Release workflow contract tests."""
-
-from pathlib import Path
-
-
-def test_release_job_uses_flattened_linux_artifact_path() -> None:
-    workflow = (
-        Path(__file__).resolve().parents[1] / ".github" / "workflows" / "publish.yaml"
-    ).read_text(encoding="utf-8")
-
-    expected = 'release-assets/agent-learning-cli-${VERSION}-linux-x64.tar.gz'
-    obsolete = (
-        'release-assets/dist-installer/'
-        'agent-learning-cli-${VERSION}-linux-x64.tar.gz'
-    )
-    assert expected in workflow
-    assert obsolete not in workflow