nisse-dev.example.yaml

@@ -25,8 +25,8 @@ llm:                                  # L9: LiteLLM is the only LLM destination
 roles:                                # L9: role -> LiteLLM model alias
   executor: qwen3.8-27b         # the local tier: best executor in the benchmark (PR #55)
   planner: claude-opus-5-5      # high-end judgement: plans a weaker executor follows
-  delegate: claude-opus-5-5     # takes over what the executor got stuck on
+  helper: claude-opus-5-5     # takes over what the executor got stuck on
-  consultant: claude-haiku-4-5  # one question at a time, no tools: cheap is enough
+  expert: claude-haiku-4-5  # one question at a time, no tools: cheap is enough
 
 matrix:                               # L5
   homeserver: https://matrix.example

nisse-dev/src/nisse_dev/agent.py

@@ -37,7 +37,7 @@ from nisse_dev.worker import SerialWorker
 # manager allows for a stop, and a hard exit mid-work is already safe.
 SHUTDOWN_WAIT_S = 10.0
 
-TOOL_DRIVING_ROLES = ("planner", "executor", "delegate")
+TOOL_DRIVING_ROLES = ("planner", "executor", "helper")
 
 
 def discover_models(config: AgentConfig, oplog: OpLog) -> ModelCatalog:
@@ -220,7 +220,7 @@ def build_agent(
     expert = ExpertDesk(
         normalized, store, config.roles, budget=tuning.expert.consult_budget
     )
-    delegate_model = config.roles.resolve("delegate")
+    delegate_model = config.roles.resolve("helper")
     delegation = Delegation(
         select_client(delegate_model, normalized, passthrough, catalog),
         store,

nisse-dev/src/nisse_dev/bench/runner.py

@@ -174,7 +174,7 @@ def run_task(
         meter = MeteringClient(client, facts)
         trims_before = getattr(client, "trimmed_requests", 0)
         model = facts.alias
-        roles = RoleMap(executor=model, planner=model, delegate=model, consultant=model)
+        roles = RoleMap(executor=model, planner=model, helper=model, expert=model)
         executor = LLMExecutor(
             meter,
             store,

nisse-dev/src/nisse_dev/config.py

@@ -36,8 +36,8 @@ REQUIRED_KEYS = (
     "llm.token",
     "roles.executor",
     "roles.planner",
-    "roles.delegate",
+    "roles.helper",
-    "roles.consultant",
+    "roles.expert",
     "matrix.homeserver",
     "matrix.user_id",
     "matrix.token",
@@ -312,8 +312,8 @@ class AgentConfig:
             roles=RoleMap(
                 executor=str(values["roles.executor"]),
                 planner=str(values["roles.planner"]),
-                delegate=str(values["roles.delegate"]),
+                helper=str(values["roles.helper"]),
-                consultant=str(values["roles.consultant"]),
+                expert=str(values["roles.expert"]),
             ),
             matrix=MatrixConfig(
                 homeserver=str(values["matrix.homeserver"]).rstrip("/"),
@@ -371,5 +371,5 @@ class AgentConfig:
             "repo": self.gitea_repo,
             "planner": self.roles.planner,
             "executor": self.roles.executor,
-            "delegate": self.roles.delegate,
+            "helper": self.roles.helper,
         }

nisse-dev/src/nisse_dev/delegation.py

@@ -184,9 +184,7 @@ class Delegation:
         # a reason, and the next step is to talk about that reason.
         self._cancel = cancel or Cancel()
 
-    def run(
+    def run(self, *, brief: str, branch: str, role: str = "helper") -> DelegationResult:
-        self, *, brief: str, branch: str, role: str = "delegate"
-    ) -> DelegationResult:
         started_at = self._now()
         base_branch = self._git.current_branch()
         # Recorded before the branch exists, so a crash anywhere after this

nisse-dev/src/nisse_dev/executor.py

@@ -165,7 +165,7 @@ class LLMExecutor:
             # that it is circling and should get on with the work or fail.
             return ToolOutcome(str(exc), success=False)
         except LLMError as exc:
-            return ToolOutcome(f"the consultant did not answer: {exc}", success=False)
+            return ToolOutcome(f"the expert did not answer: {exc}", success=False)
 
     def _scope(self) -> tuple[str | None, int]:
         """The approved plan, and where this milestone's events begin.

nisse-dev/src/nisse_dev/expert.py

@@ -17,7 +17,7 @@ memory: it survives a restart, and it is observable from outside the model,
 which is the same reasoning L9 applies to escalation triggers.
 
 Consults are recorded as ``delegation_started`` / ``delegation_completed`` with
-role ``consultant``. That is deliberate — L9 wants a "consult/delegation eval
+role ``expert``. That is deliberate — L9 wants a "consult/delegation eval
 set", and reusing the composite events gets consult evaluation for free rather
 than inventing a parallel record.
 """
@@ -35,7 +35,7 @@ from nisse_dev.canonical import (
 from nisse_dev.llm import LLMClient, LLMError, RoleMap
 from nisse_dev.state import StateStore
 
-CONSULTANT_ROLE = "consultant"
+EXPERT_ROLE = "expert"
 DEFAULT_BUDGET = 3
 
 SYSTEM_PROMPT = """You are a consultant answering one self-contained question.
@@ -111,8 +111,7 @@ class ExpertDesk:
         return sum(
             1
             for e in self._store.read_events(after=since_event_id)
-            if isinstance(e.event, DelegationStarted)
+            if isinstance(e.event, DelegationStarted) and e.event.role == EXPERT_ROLE
-            and e.event.role == CONSULTANT_ROLE
         )
 
     def ask(self, question: str, *, since_event_id: int = 0) -> ExpertAnswer:
@@ -122,11 +121,11 @@ class ExpertDesk:
                 f"consult budget of {self._budget} is spent for this task "
                 f"({used} used); escalate or ask the operator instead"
             )
-        model = self._roles.resolve(CONSULTANT_ROLE)
+        model = self._roles.resolve(EXPERT_ROLE)
         self._store.append_event(
             DelegationStarted(
                 brief=question,
-                role=CONSULTANT_ROLE,
+                role=EXPERT_ROLE,
                 model=model,
                 caps={"consults": self._budget, "used": used},
             )

nisse-dev/src/nisse_dev/llm.py

@@ -109,10 +109,10 @@ class LLMClient(Protocol):
 class RoleMap:
     executor: str
     planner: str
-    delegate: str
+    helper: str
-    consultant: str
+    expert: str
 
-    ROLES: ClassVar[tuple[str, ...]] = ("executor", "planner", "delegate", "consultant")
+    ROLES: ClassVar[tuple[str, ...]] = ("executor", "planner", "helper", "expert")
 
     def resolve(self, role: str) -> str:
         if role not in self.ROLES:

nisse-dev/tests/test_agent_build.py

@@ -263,7 +263,7 @@ def test_a_tool_less_model_warns_but_still_boots(tmp_path: Path, monkeypatch) ->
     logged = stream.getvalue()
     assert "model_tools_unsupported" in logged
     assert '"role": "executor"' in logged
-    assert '"role": "consultant"' not in logged  # ask_expert drives no tools
+    assert '"role": "expert"' not in logged  # ask_expert drives no tools
 
 
 def test_the_resolved_model_is_pinned_at_task_start(

nisse-dev/tests/test_approvals.py

@@ -20,7 +20,7 @@ from nisse_dev.planner import LLMPlanningAgent
 from nisse_dev.state import StateStore
 from nisse_dev.transport import InMemoryTransport
 
-ROLES = RoleMap(executor="x", planner="y", delegate="y", consultant="y")
+ROLES = RoleMap(executor="x", planner="y", helper="y", expert="y")
 
 
 class FakeGitea:

nisse-dev/tests/test_canonical.py

@@ -27,7 +27,7 @@ ALL_EVENTS = [
     GateOpened(gate="arch", reason="plan submitted (rev-1)"),
     DelegationStarted(
         brief="fix the flaky test",
-        role="delegate",
+        role="helper",
         model="claude-opus-5",
         caps={"turns": 20, "tokens": 100000},
     ),

nisse-dev/tests/test_catalog.py

@@ -130,13 +130,13 @@ def test_validate_names_every_bad_role_at_once() -> None:
             {
                 "executor": "local-fast",
                 "planner": "clause-opus-5",
-                "delegate": "remote-architecture",
+                "helper": "remote-architecture",
-                "consultant": "local-fast",
+                "expert": "local-fast",
             }
         )
     message = str(caught.value)
     assert "planner='clause-opus-5'" in message
-    assert "delegate='remote-architecture'" in message
+    assert "helper='remote-architecture'" in message
     assert "executor" not in message  # the good ones are not noise
 
 

nisse-dev/tests/test_config.py

@@ -16,8 +16,8 @@ COMPLETE: dict[str, Any] = {
     "roles": {
         "executor": "claude-haiku-4-5",
         "planner": "claude-opus-5",
-        "delegate": "claude-opus-5",
+        "helper": "claude-opus-5",
-        "consultant": "claude-haiku-4-5",
+        "expert": "claude-haiku-4-5",
     },
     "matrix": {
         "homeserver": "https://matrix.test/",
@@ -210,8 +210,8 @@ llm: {base_url: "https://litellm.test", token: "${TEST_LLM_TOKEN}"}
 roles:
   executor: local-fast
   planner: claude-opus-5
-  delegate: claude-opus-5
+  helper: claude-opus-5
-  consultant: claude-haiku-4-5
+  expert: claude-haiku-4-5
 matrix:
   homeserver: https://matrix.test
   user_id: "@nisse-dev-testing:test"

nisse-dev/tests/test_context.py

@@ -262,7 +262,7 @@ def test_the_executor_sees_only_its_milestone(tmp_path: Path) -> None:
             seen["system"] = system
             return [AssistantText(text="done", model=model)]
 
-    roles = RoleMap(executor="m", planner="m", delegate="m", consultant="m")
+    roles = RoleMap(executor="m", planner="m", helper="m", expert="m")
     LLMExecutor(Client(), store, roles, Workspace()).run(1)
 
     texts = [getattr(e, "text", None) for e in seen["events"]]
@@ -286,7 +286,7 @@ def test_a_request_that_cannot_fit_stops_the_milestone(tmp_path: Path) -> None:
 
     inner = RecordingClient()
     client = BudgetedClient(inner, lambda m: (4096, 1024))
-    roles = RoleMap(executor="m", planner="m", delegate="m", consultant="m")
+    roles = RoleMap(executor="m", planner="m", helper="m", expert="m")
     outcome = LLMExecutor(client, store, roles, Workspace()).run(1)
 
     assert outcome.status == "escalate"

nisse-dev/tests/test_delegation.py

@@ -385,9 +385,7 @@ def test_clean_cut_after_a_tool_result_is_resumable(
     repo.checkout_new_branch("d")
     (repo.root / "partial.py").write_text("x")
     repo.commit_all("partial work")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(ToolCall(call_id="c", tool="write_file", arguments={}))
     store.append_event(ToolResult(call_id="c", result="wrote", success=True))
     intent = pending_delegation_intent(store)
@@ -400,9 +398,7 @@ def test_dangling_tool_call_gates(
 ) -> None:
     # The effect of the call is unknown and not verifiable by inspection.
     repo.checkout_new_branch("d")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(ToolCall(call_id="c", tool="run_command", arguments={}))
     intent = pending_delegation_intent(store)
 
@@ -412,9 +408,7 @@ def test_dangling_tool_call_gates(
 def test_missing_branch_means_nothing_started(
     store: StateStore, ws: Workspace, repo: GitRepo
 ) -> None:
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(ToolResult(call_id="c", result="ok", success=True))
     intent = pending_delegation_intent(store, branch="never-created")
 
@@ -425,9 +419,7 @@ def test_a_finished_delegation_settles_rather_than_resumes(
     store: StateStore, ws: Workspace, repo: GitRepo
 ) -> None:
     repo.checkout_new_branch("d")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(
         DelegationCompleted(
             handback="done",
@@ -448,9 +440,7 @@ def test_resume_continues_the_same_delegation(
     # Interrupt after one turn, then resume: no second delegation_started,
     # because L7 defines a session as the delegation, not the process.
     repo.checkout_new_branch("d")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(ToolResult(call_id="c", result="wrote a.py", success=True))
     intent = pending_delegation_intent(store)
 
@@ -484,9 +474,7 @@ def test_recover_resumes_a_clean_cut_with_a_notice(
     store: StateStore, ws: Workspace, repo: GitRepo
 ) -> None:
     repo.checkout_new_branch("d")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(ToolResult(call_id="c", result="wrote a.py", success=True))
     pending_delegation_intent(store)
     notices: list[str] = []
@@ -505,9 +493,7 @@ def test_recover_settles_a_finished_delegation_without_resuming(
     store: StateStore, ws: Workspace, repo: GitRepo
 ) -> None:
     repo.checkout_new_branch("d")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(
         DelegationCompleted(
             handback="done",
@@ -539,9 +525,7 @@ def test_recover_leaves_an_unclean_cut_for_the_boot_gate(
     store: StateStore, ws: Workspace, repo: GitRepo
 ) -> None:
     repo.checkout_new_branch("d")
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(ToolCall(call_id="c", tool="run_command", arguments={}))
     pending_delegation_intent(store)
 

nisse-dev/tests/test_evals.py

@@ -24,7 +24,7 @@ def record_pair(
     *,
     brief: str = "add greet",
     model: str = "claude-haiku-4-5",
-    role: str = "delegate",
+    role: str = "helper",
     status: str = "handback",
 ) -> int:
     store.append_event(
@@ -56,15 +56,13 @@ def test_unpaired_start_is_skipped(store: StateStore) -> None:
     # A crash mid-delegation leaves a start with no completion; a half-formed
     # record would pollute the set L9 exists to make trustworthy.
     store.append_event(
-        DelegationStarted(brief="died", role="delegate", model="m", caps={})
+        DelegationStarted(brief="died", role="helper", model="m", caps={})
     )
     assert collect(store) == []
 
 
 def test_unrelated_events_between_the_pair_are_ignored(store: StateStore) -> None:
-    store.append_event(
+    store.append_event(DelegationStarted(brief="b", role="helper", model="m", caps={}))
-        DelegationStarted(brief="b", role="delegate", model="m", caps={})
-    )
     store.append_event(AssistantText(text="working", model="m"))
     store.append_event(
         DelegationCompleted(
@@ -97,10 +95,10 @@ def test_a_verdict_can_be_revised(store: StateStore) -> None:
 
 
 def test_consults_and_delegations_share_the_set(store: StateStore) -> None:
-    record_pair(store, role="delegate")
+    record_pair(store, role="helper")
-    record_pair(store, role="consultant", brief="which approach?")
+    record_pair(store, role="expert", brief="which approach?")
     roles = [r.role for r in collect(store)]
-    assert roles == ["delegate", "consultant"]
+    assert roles == ["helper", "expert"]
 
 
 def test_collect_can_start_after_a_marker(store: StateStore) -> None:

nisse-dev/tests/test_executor.py

@@ -20,8 +20,8 @@ from nisse_dev.tools import Workspace
 ROLES = RoleMap(
     executor="llama3.2",
     planner="claude-haiku-4-5",
-    delegate="claude-haiku-4-5",
+    helper="claude-haiku-4-5",
-    consultant="claude-haiku-4-5",
+    expert="claude-haiku-4-5",
 )
 
 

nisse-dev/tests/test_expert.py

@@ -25,8 +25,8 @@ from nisse_dev.state import StateStore
 ROLES = RoleMap(
     executor="llama3.2",
     planner="haiku",
-    delegate="haiku",
+    helper="haiku",
-    consultant="claude-haiku-4-5",
+    expert="claude-haiku-4-5",
 )
 
 GOOD = '{"answer": "Use SQLite; it is enough here.", "confidence": "high", "caveats": "Assumes one writer."}'
@@ -137,7 +137,7 @@ def test_consults_are_recorded_as_composite_events(store: StateStore) -> None:
     events = [e.event for e in store.read_events()]
     started = next(e for e in events if isinstance(e, DelegationStarted))
     completed = next(e for e in events if isinstance(e, DelegationCompleted))
-    assert started.role == "consultant"
+    assert started.role == "expert"
     assert started.brief == "q"
     assert completed.exit_status == "answered"
 

nisse-dev/tests/test_llm.py

@@ -184,8 +184,8 @@ def test_malformed_response_raises_llm_error() -> None:
 ROLES = RoleMap(
     executor="local/qwen",
     planner="anthropic/claude-opus-5",
-    delegate="anthropic/claude-opus-5",
+    helper="anthropic/claude-opus-5",
-    consultant="anthropic/claude-opus-5",
+    expert="anthropic/claude-opus-5",
 )
 
 
@@ -195,8 +195,8 @@ def test_role_map_resolves_and_pins() -> None:
     assert roles.pinned() == {
         "executor": "local/qwen",
         "planner": "anthropic/claude-opus-5",
-        "delegate": "anthropic/claude-opus-5",
+        "helper": "anthropic/claude-opus-5",
-        "consultant": "anthropic/claude-opus-5",
+        "expert": "anthropic/claude-opus-5",
     }
 
 

nisse-dev/tests/test_planner.py

@@ -29,8 +29,8 @@ from nisse_dev.transport import InMemoryTransport
 ROLES = RoleMap(
     executor="local/qwen",
     planner="anthropic/claude-opus-5",
-    delegate="anthropic/claude-opus-5",
+    helper="anthropic/claude-opus-5",
-    consultant="anthropic/claude-opus-5",
+    expert="anthropic/claude-opus-5",
 )
 
 

nisse-dev/tests/test_pr_open.py

@@ -27,8 +27,8 @@ from nisse_dev.transport import InMemoryTransport
 ROLES = RoleMap(
     executor="llama3.2",
     planner="haiku",
-    delegate="haiku",
+    helper="haiku",
-    consultant="haiku",
+    expert="haiku",
 )
 
 

nisse-dev/tests/test_review_loop.py

@@ -15,7 +15,7 @@ from nisse_dev.publish import PublishResult
 from nisse_dev.state import StateStore
 from nisse_dev.transport import InMemoryTransport
 
-ROLES = RoleMap(executor="x", planner="y", delegate="y", consultant="y")
+ROLES = RoleMap(executor="x", planner="y", helper="y", expert="y")
 
 
 class ScriptedClient:

nisse-dev/tests/test_stop.py

@@ -31,8 +31,8 @@ from nisse_dev.worker import SerialWorker
 ROLES = RoleMap(
     executor="llama3.2",
     planner="claude-haiku-4-5",
-    delegate="claude-haiku-4-5",
+    helper="claude-haiku-4-5",
-    consultant="claude-haiku-4-5",
+    expert="claude-haiku-4-5",
 )
 
 

tests/integration/nisse-test.yaml

@@ -23,8 +23,8 @@ llm:
 roles:
   executor: claude-haiku-4-5
   planner: claude-haiku-4-5
-  delegate: claude-haiku-4-5
+  helper: claude-haiku-4-5
-  consultant: claude-haiku-4-5
+  expert: claude-haiku-4-5
 
 matrix:
   homeserver: https://matrix.dongar.net

tests/integration/test_expert_live.py

@@ -22,7 +22,7 @@ pytestmark = pytest.mark.integration
 def desk(tmp_path: Path, litellm: LiteLLMClient, live_model: str):
     def make(budget: int = 2) -> tuple[ExpertDesk, StateStore]:
         model = live_model
-        roles = RoleMap(executor=model, planner=model, delegate=model, consultant=model)
+        roles = RoleMap(executor=model, planner=model, helper=model, expert=model)
         store = StateStore(
             tmp_path / "state.db", oplog=OpLog("nisse-dev", io.StringIO())
         )
@@ -43,7 +43,7 @@ def test_live_consult_parses_against_the_local_schema(desk) -> None:
     assert len(answer.answer) > 20
     # The consult is captured for L9 without any extra bookkeeping.
     [record] = collect(store)
-    assert record.role == "consultant"
+    assert record.role == "expert"
     assert record.exit_status == "answered"
 
 

tests/integration/test_milestone_live.py

@@ -64,7 +64,7 @@ def test_plan_approve_execute_produces_a_real_file(
     # Both roles use the capable model here: the point is the whole gated
     # loop over a real proxy, not local-model competence (measured separately).
     model = live_model
-    roles = RoleMap(executor=model, planner=model, delegate=model, consultant=model)
+    roles = RoleMap(executor=model, planner=model, helper=model, expert=model)
     client = litellm
     store = StateStore(tmp_path / "state.db", oplog=OpLog("nisse-dev", io.StringIO()))
     workspace = Workspace(tmp_path / "work", timeout_s=60)

tests/integration/test_planning_live.py

@@ -25,7 +25,7 @@ def test_live_planning_turn_returns_a_tool_action(
     tmp_path: Path, litellm: LiteLLMClient, live_model: str
 ) -> None:
     model = live_model
-    roles = RoleMap(executor=model, planner=model, delegate=model, consultant=model)
+    roles = RoleMap(executor=model, planner=model, helper=model, expert=model)
     store = StateStore(tmp_path / "state.db", oplog=OpLog("nisse-dev", io.StringIO()))
     store.append_event(
         UserMessage(