The call record: every model call, for the operator's study (M6.6a)

nisse-core records each model call of every agent into a file beside its
state store (<state>.calls.db), for the status page (L7, X6):

- the request body exactly as sent (never headers), the raw response with
  its reasoning, the model asked and the one that answered, time, tokens
  and llama-server's own prompt reuse (timings.cache_n);
- its work item, role, run and purpose: Driver.run and nisse-dev's
  delegation loop open runs, nested; compaction, expert consults, the
  classifier and the go-ahead reader label their calls;
- provenance, worked out in BudgetedClient by comparing the events at
  each stage: each event sent with its log id and content as sent, which
  a compaction replaced, which fitting shortened (by how much) and
  dropped.

Content is stored once by hash, compressed. It is diagnostic only: the
agent never reads it, a failed write is logged once and recording stops
while the call goes on, and the file is deleted when the agent starts,
so its layout never migrates; CallReader refuses a layout it doesn't
know. While recording, the system prompt ends with one sentence saying
so. `call_record: false` in either agent's YAML turns it off.

The benchmark records each run beside that run's own state store, and
`--keep-runs DIR` keeps both for the page.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01J5Utyfzw7gF7UhjRDEtjTk

nisse-review/src/nisse_review/build.py

@@ -9,6 +9,7 @@ from nisse_core.knowledge.memory import Memory
 from nisse_core.knowledge.vault import Vault
 from nisse_core.loop.driver import Driver
 from nisse_core.loop.expert import ExpertDesk
+from nisse_core.models import calls
 from nisse_core.models.catalog import ModelCatalog
 from nisse_core.models.compact import Compactor
 from nisse_core.models.context import BudgetedClient
@@ -94,6 +95,7 @@ def llm_client(config: ReviewConfig, oplog: OpLog | None = None) -> BudgetedClie
         safety_margin=tuning.safety_margin,
         enforces_tool_choice=lambda m: catalog.facts(m).supports_tool_choice is True,
         oplog=oplog,  # a cut that fitting has to make is logged
+        call_recorder=calls.recorder_for(config.state_db, config.call_record, oplog),
     )
 
 

nisse-review/src/nisse_review/config.py

@@ -47,6 +47,8 @@ OPTIONAL_KEYS = (
     # Patterns of the repositories it reviews ("dev/nisse", "dev/*", "*/*"),
     # public ones included; unset, every one it was given (R1, ``access``).
     "repositories",
+    # Record every model call for the operator's study (L7); on unless false.
+    "call_record",
 )
 
 
@@ -108,6 +110,7 @@ class ReviewConfig:
     gitea_operators: frozenset[str]  # whose comments are the operator's (R4)
     max_output_tokens: int = DEFAULT_MAX_OUTPUT_TOKENS
     request_timeout_s: float = DEFAULT_REQUEST_TIMEOUT_S
+    call_record: bool = True  # every model call recorded for the operator (L7)
     tuning: Tuning = field(default_factory=Tuning)
     vault: VaultConfig | None = None
     repositories: frozenset[str] | None = None  # None: all it can read
@@ -158,6 +161,9 @@ class ReviewConfig:
             owner, slash, name = pattern.partition("/")
             if not (owner and slash and name) or "/" in name:
                 errors.append(f"repositories: {pattern!r} is not owner/name")
+        record = values.get("call_record", True)
+        if not isinstance(record, bool):
+            errors.append("call_record: true or false")
         if errors:
             raise ConfigError("cannot start: " + "; ".join(errors))
         reviewer = str(values["roles.reviewer"])
@@ -173,6 +179,7 @@ class ReviewConfig:
             gitea_operators=people(values["gitea.operators"]),
             max_output_tokens=max_output,
             request_timeout_s=timeout,
+            call_record=record,
             tuning=tuning,
             vault=vault,
             repositories=patterns,

nisse-review/tests/test_review.py

@@ -645,3 +645,43 @@ def test_code_git_cannot_give_fails_the_run(tmp_path: Path) -> None:
     w["scheduler"].round()
     assert (item(w).position, pr_of(item(w)).attempts) == (READY, 1)
     assert w["gitea"].posted == []
+
+
+def test_the_review_agent_records_its_calls_unless_told_not_to(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    from nisse_core.config import ConfigError
+    from nisse_core.models.catalog import ModelCatalog
+    from nisse_review import build as review_build
+    from nisse_review.config import ReviewConfig
+
+    tree = {
+        "state_db": str(tmp_path / "s.db"),
+        "llm": {"base_url": "http://llm", "token": "k"},
+        "roles": {"reviewer": "local", "expert": "frontier"},
+        "gitea": {"url": "https://g", "token": "t", "operators": ["don"]},
+    }
+    limits = {"max_input_tokens": 32768, "max_output_tokens": 4096}
+    catalog = ModelCatalog.from_payload(
+        {
+            "data": [
+                {
+                    "model_name": m,
+                    "litellm_params": {"model": f"x/{m}"},
+                    "model_info": limits,
+                }
+                for m in ("local", "frontier")
+            ]
+        }
+    )
+    monkeypatch.setattr(review_build.ModelCatalog, "discover", lambda *a, **k: catalog)
+
+    on = review_build.llm_client(ReviewConfig.from_tree(tree, {}))
+    assert on.call_recorder is not None
+    assert on.call_recorder.path == tmp_path / "s.calls.db"
+    off = review_build.llm_client(
+        ReviewConfig.from_tree({**tree, "call_record": False}, {})
+    )
+    assert off.call_recorder is None
+    with pytest.raises(ConfigError, match="call_record"):
+        ReviewConfig.from_tree({**tree, "call_record": "yes"}, {})