agentx-python 0.8.8__py3-none-any.whl → 0.8.10__py3-none-any.whl

agentx/cli.py

@@ -4,18 +4,21 @@ Evaluate, Monitor), published separately at github.com/AgentX-ai/AgentX-trace-ev
 wrapping a Bun-compiled TypeScript engine, not Python). That compiled engine binary is tens of
 megabytes; most `pip install agentx-python` installs are just this SDK talking to the hosted
 AgentX SaaS and would never touch it, so it isn't bundled in this package. Instead, this command
-downloads the matching release into ~/.agentx/bin the first time it's needed (mirroring
+downloads the release into ~/.agentx/bin the first time it's needed (mirroring
 AgentX-trace-eval's own install.sh) and then hands off to the real `agentx-server` binary.
 
-The installed release tag is stamped in ~/.agentx/bin/.version. The launcher never silently
-re-downloads, but it re-installs when you ask: `--update` fetches the newest release (engine +
-dashboard), and setting AGENTX_TRACE_EVAL_VERSION to a tag other than the stamped one switches
-to that release. When running "latest", a quick fail-open check against GitHub tells you if a
-newer release exists.
+Versioning: each SDK release pins the engine release it was tested against
+(agentx.version.ENGINE_VERSION); the launcher installs that pin and converges to it - if the
+stamped install (~/.agentx/bin/.version) differs from the pin, it re-installs, so upgrading the
+SDK upgrades the engine to the matching pair. `--update` force-reinstalls the resolved version
+(recovering a broken or pre-stamp install). Set AGENTX_TRACE_EVAL_VERSION to a tag or `latest`
+to override the pin; in `latest` mode the launcher trusts whatever is installed, prints a
+fail-open notice when GitHub has something newer, and `--update` fetches it.
 
 Usage:
     agentx-trace-eval --dev
     agentx-trace-eval --update --dev
+    AGENTX_TRACE_EVAL_VERSION=latest agentx-trace-eval --update --dev
     agentx-trace-eval --port 5000 --db-url postgres://...
 """
 
@@ -32,6 +35,8 @@ from typing import Optional, Tuple
 
 import requests
 
+from agentx.version import ENGINE_VERSION
+
 REPO = "AgentX-ai/AgentX-trace-eval"
 INSTALL_DIR = Path(os.environ.get("AGENTX_INSTALL_DIR", str(Path.home() / ".agentx" / "bin")))
 _BIN_NAMES = ("agentx", "agentx-server", "agentx-engine")
@@ -114,11 +119,17 @@ def _latest_release_tag() -> Optional[str]:
         return None
 
 
+def resolve_version() -> str:
+    """The engine version this launcher should be running: the AGENTX_TRACE_EVAL_VERSION
+    override (a tag, or `latest`) when set, else the SDK's tested pin."""
+    return os.environ.get("AGENTX_TRACE_EVAL_VERSION") or ENGINE_VERSION
+
+
 def _install(version: str = "latest") -> None:
     os_name, arch = _platform_tag()
     INSTALL_DIR.mkdir(parents=True, exist_ok=True)
 
-    print(f"agentx-trace-eval: downloading agentx ({os_name}/{arch})...", file=sys.stderr)
+    print(f"agentx-trace-eval: downloading agentx {version} ({os_name}/{arch})...", file=sys.stderr)
     with tempfile.TemporaryDirectory() as tmp:
         tmp_dir = Path(tmp)
         archive = tmp_dir / "agentx.tar.gz"
@@ -178,23 +189,30 @@ def _install(version: str = "latest") -> None:
         _extract_tar(web_archive, web_dir)
 
 
-def ensure_installed(version: str = "latest", force: bool = False) -> Path:
-    """Downloads agentx-server (+ its engine and dashboard) into ~/.agentx/bin when it's missing
+def ensure_installed(version: Optional[str] = None, force: bool = False) -> Path:
+    """Installs agentx-server (+ its engine and dashboard) into ~/.agentx/bin when it's missing
     there, when `force` is set, or when `version` is a specific tag that differs from the
-    installed one. Returns the path to the agentx-server executable. Set AGENTX_INSTALL_DIR to
-    change where this looks/installs; set AGENTX_TRACE_EVAL_VERSION to pin a release tag."""
+    stamped install - the convergence that makes an SDK upgrade carry its tested engine along.
+    `version` defaults to resolve_version() (the AGENTX_TRACE_EVAL_VERSION override, else the
+    SDK's pin); pass "latest" to trust whatever is installed and only fetch when missing.
+    Returns the path to the agentx-server executable. Set AGENTX_INSTALL_DIR to change where
+    this looks/installs."""
+    if version is None:
+        version = resolve_version()
     server_path = INSTALL_DIR / "agentx-server"
-    installed = _stamped_version()
-    pin_changed = version != "latest" and installed is not None and installed != version
-    if force or pin_changed or not server_path.exists():
+    stamped = _stamped_version()
+    pin_mismatch = version != "latest" and stamped != version
+    if force or pin_mismatch or not server_path.exists():
         _install(version=version)
     return server_path
 
 
 def _maybe_print_update_notice(installed: Optional[str]) -> None:
+    """Only relevant in `latest` mode, where nothing converges automatically - a quick fail-open
+    check tells the user when the world has moved on. (Pinned mode needs no notice: a pin
+    mismatch re-installs instead of nagging.)"""
     latest = _latest_release_tag()
     if installed is None:
-        # Pre-stamp install (or a wiped stamp): age unknown, so always point at --update.
         print(
             "agentx-trace-eval: installed engine version unknown"
             + (f" (newest release is {latest})" if latest else "")
@@ -215,7 +233,7 @@ def main() -> None:
     # --update belongs to this launcher, not to agentx-server - consume it before the handoff.
     args = [a for a in args if a != "--update"]
 
-    version = os.environ.get("AGENTX_TRACE_EVAL_VERSION", "latest")
+    version = resolve_version()
     server_path = ensure_installed(version=version, force=force_update)
     if not server_path.exists():
         raise SystemExit(f"agentx-trace-eval: {server_path} still missing after install, giving up")

agentx/evaluations/client.py

@@ -37,6 +37,9 @@ _RETRY_BACKOFF = [1.0, 2.0, 4.0]
 # wait out the whole job on one connection. Matches EvaluationRunContext.analyze()'s own
 # default timeout.
 _SELF_HOST_ANALYZE_TIMEOUT = 1800
+# Batch result submission scores each result synchronously inside the request (one judge call
+# per result on the sync path) - a big batch on a slow judge legitimately takes minutes.
+_SELF_HOST_SCORING_TIMEOUT = 900
 
 
 class AgentXEvaluationsError(Exception):
@@ -344,7 +347,15 @@ class EvaluationsClient:
             "batchId": batch_id,
             "results": [_result_to_payload(r) for r in results],
         }
-        data = self._request("POST", f"/runs/{run_id}/results", json=payload)
+        # Scoring is synchronous inside this request (a judge call per result - the runner's
+        # spinner says "~60s+" for a reason), so the default 30s timeout + silent backoff loop
+        # re-POSTed the batch WHILE the first submission was still scoring, and the two raced
+        # into the engine's (run_id, idempotency_key) unique constraint. Long timeout, no
+        # transport retry - the runner's own logged retry-once is the retry layer here.
+        data = self._request(
+            "POST", f"/runs/{run_id}/results", json=payload,
+            timeout=_SELF_HOST_SCORING_TIMEOUT, retry=False,
+        )
         return BatchAppendResponse(**data)
 
     def finalize_run(self, run_id: str) -> Dict[str, Any]:

agentx/version.py

@@ -1 +1,7 @@
-VERSION = "0.8.8"
+VERSION = "0.8.10"
+
+# The AgentX-trace-eval release this SDK version is tested against - what `agentx-trace-eval`
+# installs and converges to (see agentx/cli.py). Bump together with VERSION when releasing, so
+# every published SDK names a known-good engine+dashboard pair. Users can override with
+# AGENTX_TRACE_EVAL_VERSION=<tag|latest>.
+ENGINE_VERSION = "v0.3.1"

{agentx_python-0.8.8.dist-info → agentx_python-0.8.10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentx-python
-Version: 0.8.8
+Version: 0.8.10
 Summary: Official Python SDK for AgentX (https://www.agentx.so/)
 Home-page: https://github.com/AgentX-ai/AgentX-python
 Author: Robin Wang and AgentX Team
@@ -193,7 +193,7 @@ client.evaluations.run(
 
 See [Prompt registry](EVALUATIONS.md#prompt-registry) in the full guide, or [self-host's docs](https://docs.agentx.so/improve/prompt-management) for the "Suggest improvement" dashboard flow (self-host only - no hosted-SaaS equivalent yet).
 
-On self-host, a finalized run can also **gate a CI job**: `report.gate(fail_under=7, no_regression=True)` checks the run's average rating against an absolute floor and/or the dataset's previous run, prints per-check verdicts into the CI log, and returns an exit code - `sys.exit(gate.exit_code)` blocks the merge on regression. Recorded gates appear in the dashboard's CI Gates tab. See [self-host's CI docs](https://docs.agentx.so/integrations/self-host-ci) for the GitHub Actions recipe.
+On self-host, a finalized run can also **gate a CI job**: `run.gate(fail_under=7, no_regression=True)` (on the run context `.execute()` returns) checks the run's average rating against an absolute floor and/or the dataset's previous run, prints per-check verdicts into the CI log, and returns an exit code - `sys.exit(gate.exit_code)` blocks the merge on regression. Recorded gates appear in the dashboard's CI Gates tab. See [self-host's CI docs](https://docs.agentx.so/integrations/self-host-ci) for the GitHub Actions recipe.
 
 See **[EVALUATIONS.md](EVALUATIONS.md)** for the full guide - dataset builder, framework adapters, similarity metrics, smoke testing, judge configuration, prompt registry, and the complete API reference.
 

{agentx_python-0.8.8.dist-info → agentx_python-0.8.10.dist-info}/RECORD

@@ -1,6 +1,6 @@
 agentx/__init__.py,sha256=4iMiGLNU4S-I-WyX0Z7l0BFUY_krVJ3seMQPqyO-ReA,602
 agentx/agentx.py,sha256=WuzVQtXSPqBhRwKxkUlQ8VG2cwvGpbMcti4n6aTNPwA,8816
-agentx/cli.py,sha256=4c6hPaz5wCvG-IOvzwqlO3PRpkWK4ocf9K-6UXbEfbU,9504
+agentx/cli.py,sha256=yawLQLSeZ7KIl7ukgmly-lrYbx9GgKcaI9M4xamiMPw,10468
 agentx/exceptions.py,sha256=2tDLdZjriFBQbjgzyIcah12RazpbujhqZSRbSXgAml4,1377
 agentx/export.py,sha256=FNT00e-j9AQRkNDksUwgHJXgr8-XVPscEH4xUqJHmfM,4265
 agentx/feedback.py,sha256=hMPU3ivUHvhOXOH7gvT4Jy1yztVmoIMOw1MbUNRs5Gk,3643
@@ -10,10 +10,10 @@ agentx/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 agentx/testing.py,sha256=0shZEid_vhJgBegpO_loUq6APYDxQrUgD3jvcRdIDV8,7372
 agentx/traces.py,sha256=sz9gxutlDKNgf0fdCkyzKsGwK5eMAJuIEl3NIwGrsuo,2287
 agentx/util.py,sha256=lt2Kpg4Fj8whrSuAhLEahcPEIYRVpK9RCC5y9ShdTJs,703
-agentx/version.py,sha256=KtsrDPnnfwQ504ziRmPTLeUxADR8tN7iC1IRjn_cl_M,18
+agentx/version.py,sha256=sSbsfDvIXz3zR_dv_mmgv3XPzX8Fi8FnwzRLQ-G08M8,365
 agentx/evaluations/__init__.py,sha256=Erv7RGFlRxGTG4rVb2uHhCqLLEX_iAWKqkZKzK6CumE,262
 agentx/evaluations/_term.py,sha256=WFpiNzdgDBeJJ-Gg-6X7TwwxllobuE4OqFUTuDQvS3Y,2529
-agentx/evaluations/client.py,sha256=miUmEq4IthUDl-QVAczqT7iaOu4X6QMbHhlm2rSSJpg,31239
+agentx/evaluations/client.py,sha256=UlYl8OFmIjk8oqmL_LEpjnTn6CmUPBM1kgVgzkUKfT8,31999
 agentx/evaluations/datasets.py,sha256=2--V40JrSJoXxR1s0jI3i5rYT8ys25ouijqQQlj72D8,16053
 agentx/evaluations/evaluation_settings.py,sha256=geRydHqKpxwkxBVfMIOqkB6NlStyOZrYwZnMHluDwvA,6276
 agentx/evaluations/models.py,sha256=fKaDwyXhruXRHpBDw89Vv_FeceLvSILVrA9HNtzKLF8,32266
@@ -63,9 +63,9 @@ agentx/tracing/ci_types.py,sha256=zFVcZGvc1qbVDdOlEPQ1i5R6terqJHxCzFF4ZjouxxI,14
 agentx/tracing/eval_scope.py,sha256=ElMbPxpqpQBVuaUnHNlw9RIQu71oyOpd0yR55bDH8eI,2144
 agentx/tracing/ingest_client.py,sha256=5X6RyxtMq1XYgOzsMd3b2f2qpmYpsMnhKm7NB3oup00,17664
 agentx/tracing/tracer.py,sha256=qQojU-T4O8IJKyf9vue0Em66qPBTZKl3FMyr3KJKPnI,53770
-agentx_python-0.8.8.dist-info/licenses/LICENSE,sha256=gZVsM-nLsE8vlaY6NXXsVoo6IlCClxkToAWmhgT3y_s,10762
-agentx_python-0.8.8.dist-info/METADATA,sha256=554KW0UbN4YVvfAtI4sudXO-2Dscr0biu4Q2W8VxaHg,20992
-agentx_python-0.8.8.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
-agentx_python-0.8.8.dist-info/entry_points.txt,sha256=rQqF1JTY3T1yfviBU1r8PU-5mBdfOk6P4bZ1L4BskIM,172
-agentx_python-0.8.8.dist-info/top_level.txt,sha256=s-q-HB9Gb_QdrZNacSeQyF_c25gQooMy7DlxzgLOHPk,7
-agentx_python-0.8.8.dist-info/RECORD,,
+agentx_python-0.8.10.dist-info/licenses/LICENSE,sha256=gZVsM-nLsE8vlaY6NXXsVoo6IlCClxkToAWmhgT3y_s,10762
+agentx_python-0.8.10.dist-info/METADATA,sha256=EzbfWsvWLIimWsfA-vHWsqV0BWRK4Emc1u1ryY_xWHU,21032
+agentx_python-0.8.10.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+agentx_python-0.8.10.dist-info/entry_points.txt,sha256=rQqF1JTY3T1yfviBU1r8PU-5mBdfOk6P4bZ1L4BskIM,172
+agentx_python-0.8.10.dist-info/top_level.txt,sha256=s-q-HB9Gb_QdrZNacSeQyF_c25gQooMy7DlxzgLOHPk,7
+agentx_python-0.8.10.dist-info/RECORD,,