nodloop 0.10.0__tar.gz → 0.10.1__tar.gz

{nodloop-0.10.0 → nodloop-0.10.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: nodloop
-Version: 0.10.0
+Version: 0.10.1
 Summary: Record AI runs and the verdicts people give on them, and hand the knowledge they approve to the next run in the same place
 Project-URL: Homepage, https://github.com/jeon-jihyeon/nodloop
 Project-URL: Repository, https://github.com/jeon-jihyeon/nodloop

{nodloop-0.10.0 → nodloop-0.10.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "nodloop"
-version = "0.10.0"
+version = "0.10.1"
 description = "Record AI runs and the verdicts people give on them, and hand the knowledge they approve to the next run in the same place"
 readme = "README.md"
 license = "MIT"

{nodloop-0.10.0 → nodloop-0.10.1}/src/nodloop/client.py

@@ -33,7 +33,7 @@ __all__ = [
 ]
 
 # The binary release the client fetches when no nodloop is installed
-VERSION = "0.10.0"
+VERSION = "0.10.1"
 
 
 # The reports report reads
@@ -330,15 +330,26 @@ class Client:
             args["edited_output"] = edited
         await self.call("feedback", args)
 
-    async def outcome(self, run: str, result: Result, confirmed_cause: str = "", note: str = "", reviewer: str = "") -> None:
+    async def outcome(
+        self,
+        run: str,
+        result: Result,
+        confirmed_cause: str = "",
+        note: str = "",
+        reviewer: str = "",
+        *,
+        actor: Literal["person", "code", "model", "unknown"] | None = None,
+    ) -> None:
         """Records what a real check found for a run: confirmed, refuted or inconclusive
 
         A different fact from the verdict on the output
         confirmed_cause only goes with confirmed
         reviewer names who checked and is required over stdio while a server key names itself
+        actor names what kind of evaluator checked
+        It never proves a person since only the transport does
         """
         args: dict[str, Any] = {"trace_id": run, "result": result}
-        for key, value in (("confirmed_cause", confirmed_cause), ("note", note), ("reviewer", reviewer)):
+        for key, value in (("confirmed_cause", confirmed_cause), ("note", note), ("reviewer", reviewer), ("actor", actor)):
             if value:
                 args[key] = value
         await self.call("outcome", args)

{nodloop-0.10.0 → nodloop-0.10.1}/src/nodloop/records.py

@@ -244,6 +244,7 @@ class _ExtractRecordOutputAttemptsDraftRequired(TypedDict):
 class ExtractRecordOutputAttemptsDraft(_ExtractRecordOutputAttemptsDraftRequired, total=False):
     relates_to: str
     keys: list[str] | None
+    excerpt: str
 
 
 class ExtractRecordOutputAttemptsCritique(TypedDict):
@@ -273,6 +274,22 @@ class ExtractRecordOutputRelated(TypedDict):
     status: Literal["candidate", "approved", "retired", "superseded"]
 
 
+class _ExtractRecordOutputTriageRequired(TypedDict):
+    policy: str
+    nature: Literal["preference", "safety", "organization", "unknown"]
+    mixed: bool
+    reversibility: Literal["H", "M", "L", "unknown"]
+    source: Literal["complete", "incomplete", "missing", "scope"]
+    protected: Literal["none", "safety", "organization", "mixed", "conflict", "artifact", "unknown"]
+    lasting: bool
+    faithful: bool
+    conflict: bool
+
+
+class ExtractRecordOutputTriage(_ExtractRecordOutputTriageRequired, total=False):
+    classify: str
+
+
 class _ExtractRecordOutputRequired(TypedDict):
     attempts: list[ExtractRecordOutputAttempts] | None
     conclusion: str
@@ -281,6 +298,8 @@ class _ExtractRecordOutputRequired(TypedDict):
 class ExtractRecordOutput(_ExtractRecordOutputRequired, total=False):
     candidate: ExtractRecordOutputCandidate | None
     related: ExtractRecordOutputRelated | None
+    excerpt: str
+    triage: ExtractRecordOutputTriage | None
 
 
 class _ExtractRecordUsageRequired(TypedDict):
@@ -526,6 +545,15 @@ class RefillRecord(_RefillRecordRequired, total=False):
     labels: dict[str, list[str] | None]
 
 
+class TurnRecordInputContext(TypedDict):
+    id: str
+    version: int
+
+
+class TurnRecordInput(TypedDict, total=False):
+    context: list[TurnRecordInputContext] | None
+
+
 class _TurnRecordUsageRequired(TypedDict):
     input_tokens: int
     output_tokens: int
@@ -544,14 +572,14 @@ class _TurnRecordRequired(TypedDict):
     id: str
     name: Literal["turn"]
     time: str
-    input: dict[str, Any]
+    input: TurnRecordInput
     usage: TurnRecordUsage
     duration_ms: int
     schema_version: Literal[1]
 
 
 class TurnRecord(_TurnRecordRequired, total=False):
-    """The start of a turn of a session"""
+    """The start of a turn of a session with its context item references"""
 
     session_id: str
     subject: str
@@ -741,9 +769,26 @@ class DecideRecordInput(_DecideRecordInputRequired, total=False):
     proposal: str
 
 
+class _DecideRecordOutputTriageRequired(TypedDict):
+    policy: str
+    nature: Literal["preference", "safety", "organization", "unknown"]
+    mixed: bool
+    reversibility: Literal["H", "M", "L", "unknown"]
+    source: Literal["complete", "incomplete", "missing", "scope"]
+    protected: Literal["none", "safety", "organization", "mixed", "conflict", "artifact", "unknown"]
+    lasting: bool
+    faithful: bool
+    conflict: bool
+
+
+class DecideRecordOutputTriage(_DecideRecordOutputTriageRequired, total=False):
+    classify: str
+    grant: str
+
+
 class _DecideRecordOutputRequired(TypedDict):
     tier: int
-    action: Literal["apply", "withdraw", "drop", "join", "ask", "shadow", "retry"]
+    action: Literal["apply", "withdraw", "drop", "join", "ask", "shadow", "retry", "hold"]
 
 
 class DecideRecordOutput(_DecideRecordOutputRequired, total=False):
@@ -763,6 +808,8 @@ class DecideRecordOutput(_DecideRecordOutputRequired, total=False):
         "judge_against",
         "judge_undecided",
         "judge_split",
+        "delegated",
+        "triage_protected",
     ]
     judge: str
     outcome: Literal["conform", "against", "undecided", "split"]
@@ -772,6 +819,8 @@ class DecideRecordOutput(_DecideRecordOutputRequired, total=False):
     reason: Literal["duplicate", "no_evidence", "temp_scope", "conflict", "words", "compaction", "person"]
     failure: Literal["hard", "repairable", "underexplored", "unpromising"]
     reopen: Literal["other_session", "permanent_place", "next_pass", "person_only"]
+    triage: DecideRecordOutputTriage | None
+    delegation: str
 
 
 class _DecideRecordUsageRequired(TypedDict):
@@ -808,7 +857,7 @@ class _DecideRecordRequired(TypedDict):
     input: DecideRecordInput
     usage: DecideRecordUsage
     duration_ms: int
-    schema_version: Literal[1]
+    schema_version: Literal[2]
 
 
 class DecideRecord(_DecideRecordRequired, total=False):
@@ -862,6 +911,7 @@ class KnobRecordInputKnobs(_KnobRecordInputKnobsRequired, total=False):
     audit: KnobRecordInputKnobsAudit
     target: KnobRecordInputKnobsTarget
     cusum: KnobRecordInputKnobsCusum
+    delegation_paused: list[str] | None
 
 
 class KnobRecordInputJudgeSets(TypedDict):
@@ -878,8 +928,10 @@ class _KnobRecordInputJudgeRequired(TypedDict):
 
 
 class KnobRecordInputJudge(_KnobRecordInputJudgeRequired, total=False):
+    backend: str
     kept: int
     refused: int
+    answered: str
 
 
 class _KnobRecordInputChangesOpeningRequired(TypedDict):
@@ -891,7 +943,7 @@ class KnobRecordInputChangesOpening(_KnobRecordInputChangesOpeningRequired, tota
 
 
 class _KnobRecordInputChangesRequired(TypedDict):
-    rule: Literal["posterior", "alarm", "raise", "lower", "person"]
+    rule: Literal["posterior", "alarm", "raise", "lower", "person", "delegation"]
 
 
 class KnobRecordInputChanges(_KnobRecordInputChangesRequired, total=False):
@@ -900,6 +952,7 @@ class KnobRecordInputChanges(_KnobRecordInputChangesRequired, total=False):
     differ: int
     p: float
     knob: str
+    grant: str
 
 
 class _KnobRecordInputRequired(TypedDict):
@@ -938,7 +991,7 @@ class _KnobRecordRequired(TypedDict):
     input: KnobRecordInput
     usage: KnobRecordUsage
     duration_ms: int
-    schema_version: Literal[1]
+    schema_version: Literal[2]
 
 
 class KnobRecord(_KnobRecordRequired, total=False):
@@ -1384,6 +1437,8 @@ class OutcomeRecord(_OutcomeRecordRequired, total=False):
 
     confirmed_cause: str
     note: str
+    actor: Literal["person", "code", "model", "unknown"]
+    proof: Literal["key", "terminal", "elicitation", "reply", "none"]
 
 
 class _KnowledgeRecordRunRequired(TypedDict):

{nodloop-0.10.0 → nodloop-0.10.1}/tests/test_client.py

@@ -304,13 +304,18 @@ async def test_outcome_and_report(binary: str, records: str) -> None:
     (Path(records) / "prune.json").write_text('{"at":"2020-01-02T00:00:00Z","cutoff":"2020-01-01T00:00:00Z"}')
     async with Client(records, binary) as c:
         run = await c.record("stream-bot", "lag comes from partition 7", labels=ACME)
-        await c.outcome(run, "confirmed", confirmed_cause="partition 7 was hot", note="checked the broker", reviewer="ann")
+        await c.outcome(run, "confirmed", confirmed_cause="partition 7 was hot", note="checked the broker", reviewer="ann", actor="code")
+        outcomes = await c.records("outcome")
+        assert outcomes.records[0]["actor"] == "code"
+        assert "proof" not in outcomes.records[0]
         health = await c.report("health")
         loop = await c.report("loop")
         with pytest.raises(ToolError):
             await c.outcome(run, "refuted", confirmed_cause="partition 7 was hot", reviewer="ann")
         with pytest.raises(ToolError, match="a reviewer is required"):
             await c.outcome(run, "refuted")
+        with pytest.raises(ToolError, match="actor"):
+            await c.call("outcome", {"trace_id": run, "result": "confirmed", "reviewer": "ann", "actor": "future"})
 
     assert isinstance(health.report, (list, dict)) and loop.report["totals"]["runs"] == 1
     assert loop.since == "2020-01-01T00:00:00Z"

{nodloop-0.10.0 → nodloop-0.10.1}/uv.lock

@@ -900,7 +900,7 @@ wheels = [
 
 [[package]]
 name = "nodloop"
-version = "0.10.0"
+version = "0.10.1"
 source = { editable = "." }
 dependencies = [
     { name = "mcp" },