M2.8: switch over to the new loop

nisse-dev now runs only the conversational loop. The gated state machine
it was built alongside is gone:

- Removed control.py, planner.py, executor.py, approvals.py, gates.py
  (command_form moves to loop/reader.py), publish.py and poller.py, the
  old assembly in agent.py (its shared helpers move to loop/build.py),
  the store's state-machine view (get_state, transition), their tests,
  the old end-to-end and live planning/milestone/gate tests, and
  specs/orchestrator.qnt with its scenarios.
- Roles are driver, helper, expert and classifier; driver is required.
  loop:, roles.planner, roles.executor and tuning.executor are refused
  with what to use instead; tuning.driver takes the turn cap and failure
  limit.
- The benchmark and the capability matrix measure the driver: the
  loop's own system prompt and tools, the task's plan as the agreed
  restatement. Task plans end with finish_milestone, so earlier runs
  show as (old); stored runs' executor tuning reads as the driver's.
- A delegation on the loop records what its helper's alias resolved to
  (L9), as the old path's did.
- The dump shows a work item's revision and what it waits for.
- Behaviour the old tests covered and the loop shares stays covered:
  driver runs that fail on a model call, stop mid-turn, reset the
  failure count; go-aheads ignoring case and punctuation; a request too
  big to fit is never sent; the assembly tests, now on build_loop_agent.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPQ14L3LikyefCMFBqEbKE

nisse-dev/tests/test_config.py

@@ -4,7 +4,7 @@ from typing import Any
 
 import pytest
 
-from nisse_dev.config import AgentConfig, ConfigError
+from nisse_dev.config import AgentConfig, ConfigError, Project
 from nisse_dev.llm import DEFAULT_REQUEST_TIMEOUT_S
 
 # A complete config as the YAML file would give it, with secrets by reference.
@@ -19,9 +19,8 @@ COMPLETE: dict[str, Any] = {
     },
     "llm": {"base_url": "https://litellm.test", "token": "${TEST_LLM_TOKEN}"},
     "roles": {
-        "executor": "claude-haiku-4-5",
-        "planner": "claude-opus-5",
-        "helper": "claude-opus-5",
+        "driver": "claude-opus-5",
+        "helper": "claude-haiku-4-5",
         "expert": "claude-haiku-4-5",
     },
     "matrix": {
@@ -64,22 +63,27 @@ def build(**changes: Any) -> AgentConfig:
     return AgentConfig.from_tree(tree(**changes), SECRETS)
 
 
+def only(config: AgentConfig) -> Project:
+    [project] = config.projects
+    return project
+
+
 def test_complete_config_builds_with_secrets_by_reference() -> None:
     config = build()
-    assert config.project == "greet"
+    assert only(config).name == "greet"
     assert config.llm_token == "sk-x"
     assert config.gitea_token == "gt"
-    assert config.roles.planner == "claude-opus-5"
+    assert config.roles.driver == "claude-opus-5"
     assert config.matrix.room_id == "!room:test"
     assert config.matrix.homeserver == "https://matrix.test"
-    assert config.gitea_repo == "don/demo"
-    assert config.base_branch == "main"
-    assert config.tuning.executor.max_turns == 40  # absent section: defaults
+    assert only(config).repo == "don/demo"
+    assert only(config).base_branch == "main"
+    assert config.tuning.driver.max_turns == 40  # absent section: defaults
 
 
 def test_paths_expand_the_home_directory() -> None:
     config = build()
-    assert config.workspace.is_absolute()
+    assert only(config).workspace.is_absolute()
     assert "~" not in str(config.state_db)
 
 
@@ -119,7 +123,7 @@ def test_an_operator_list_of_only_blanks_is_missing() -> None:
 
 
 def test_base_branch_is_configurable() -> None:
-    assert build(**{f"{P}base_branch": "trunk"}).base_branch == "trunk"
+    assert only(build(**{f"{P}base_branch": "trunk"})).base_branch == "trunk"
 
 
 def test_max_output_tokens_is_configurable_and_checked() -> None:
@@ -137,8 +141,8 @@ def test_request_timeout_is_configurable_and_checked() -> None:
 
 
 def test_the_request_timeout_reaches_both_clients() -> None:
-    from nisse_dev.agent import make_llm_clients
     from nisse_dev.catalog import ModelCatalog
+    from nisse_dev.loop.build import make_llm_clients
 
     normalized, passthrough = make_llm_clients(
         build(llm__request_timeout_s=42), ModelCatalog({})
@@ -158,7 +162,7 @@ def test_redacted_summary_carries_no_secrets() -> None:
 def test_tuning_is_read() -> None:
     config = build(
         tuning={
-            "executor": {"max_turns": 40, "failure_limit": 3},
+            "driver": {"max_turns": 40, "failure_limit": 3},
             "expert": {"consult_budget": 5},
             "delegation": {"max_turns": 20, "max_tokens": 100000, "wall_clock_s": 600},
             "tools": {"command_timeout_s": 300, "max_output_chars": 2000},
@@ -166,7 +170,7 @@ def test_tuning_is_read() -> None:
         }
     )
     tuning = config.tuning
-    assert (tuning.executor.max_turns, tuning.executor.failure_limit) == (40, 3)
+    assert (tuning.driver.max_turns, tuning.driver.failure_limit) == (40, 3)
     assert tuning.expert.consult_budget == 5
     assert tuning.delegation.max_tokens == 100_000
     assert tuning.delegation.wall_clock_s == 600.0
@@ -175,8 +179,8 @@ def test_tuning_is_read() -> None:
 
 
 def test_a_partly_given_tuning_section_keeps_the_other_defaults() -> None:
-    tuning = build(tuning={"executor": {"max_turns": 40}}).tuning
-    assert tuning.executor.failure_limit == 2
+    tuning = build(tuning={"driver": {"max_turns": 40}}).tuning
+    assert tuning.driver.failure_limit == 2
 
 
 def test_every_problem_is_reported_at_once() -> None:
@@ -184,7 +188,7 @@ def test_every_problem_is_reported_at_once() -> None:
     # with no effect, so unknown keys are errors.
     bad = tree(llm_extra=1, **{f"{P}workspace": None})
     bad["tuning"] = {
-        "executor": {"max_turn": 40, "failure_limit": 0},
+        "driver": {"max_turn": 40, "failure_limit": 0},
         "expert": {"consult_budget": True},
         "bogus": {"x": 1},
     }
@@ -194,8 +198,8 @@ def test_every_problem_is_reported_at_once() -> None:
     for expected in (
         "projects.greet: unset or empty — workspace",
         "llm_extra: unknown setting",
-        "tuning.executor.max_turn: unknown setting",
-        "tuning.executor.failure_limit: must be a positive number",
+        "tuning.driver.max_turn: unknown setting",
+        "tuning.driver.failure_limit: must be a positive number",
         "tuning.expert.consult_budget: must be a positive number",
         "tuning.bogus: unknown section",
     ):
@@ -203,8 +207,8 @@ def test_every_problem_is_reported_at_once() -> None:
 
 
 def test_a_whole_number_is_required_where_one_is_meant() -> None:
-    with pytest.raises(ConfigError, match="tuning.executor.max_turns"):
-        build(tuning={"executor": {"max_turns": 2.5}})
+    with pytest.raises(ConfigError, match="tuning.driver.max_turns"):
+        build(tuning={"driver": {"max_turns": 2.5}})
 
 
 # --- the file --------------------------------------------------------------
@@ -220,8 +224,7 @@ projects:
       key: ~/.ssh/nisse-dev
 llm: {base_url: "https://litellm.test", token: "${TEST_LLM_TOKEN}"}
 roles:
-  executor: local-fast
-  planner: claude-opus-5
+  driver: local-fast
   helper: claude-opus-5
   expert: claude-haiku-4-5
 matrix:
@@ -245,7 +248,7 @@ def write(directory: Path, text: str = FILE) -> Path:
 
 def test_a_file_is_read(tmp_path: Path) -> None:
     config = AgentConfig.from_file(write(tmp_path), SECRETS)
-    assert config.roles.executor == "local-fast"
+    assert config.roles.driver == "local-fast"
     assert config.config_file == tmp_path / "nisse-dev.yaml"
 
 
@@ -256,7 +259,9 @@ def test_invalid_yaml_is_reported(tmp_path: Path) -> None:
 
 def test_load_uses_nisse_config(tmp_path: Path) -> None:
     path = write(tmp_path)
-    assert AgentConfig.load({**SECRETS, "NISSE_CONFIG": str(path)}).project == "greet"
+    assert (
+        only(AgentConfig.load({**SECRETS, "NISSE_CONFIG": str(path)})).name == "greet"
+    )
 
 
 def test_load_defaults_to_the_working_directory(
@@ -264,7 +269,7 @@ def test_load_defaults_to_the_working_directory(
 ) -> None:
     write(tmp_path)
     monkeypatch.chdir(tmp_path)
-    assert AgentConfig.load(SECRETS).project == "greet"
+    assert only(AgentConfig.load(SECRETS)).name == "greet"
 
 
 def test_load_without_a_file_says_how_to_make_one(
@@ -284,8 +289,8 @@ def test_the_example_file_is_valid() -> None:
         "NISSE_GITEA_TOKEN": "gt",
     }
     config = AgentConfig.from_file(example, env)
-    assert config.tuning.executor.max_turns == 40
-    assert config.roles.executor == "qwen3.8-27b"
+    assert config.tuning.driver.max_turns == 40
+    assert config.roles.driver == "claude-opus-5-5"
 
 
 # --- nothing that carries secrets may sit inside the clone -----------------
@@ -294,7 +299,7 @@ def test_the_example_file_is_valid() -> None:
 def test_state_db_and_workspace_are_separate() -> None:
     # L8: build artefacts must not share a volume with agent state.
     config = build()
-    assert config.state_db.parent != config.workspace
+    assert config.state_db.parent != only(config).workspace
 
 
 def test_state_db_inside_the_clone_is_refused(tmp_path: Path) -> None:
@@ -333,7 +338,7 @@ def test_sibling_paths_are_fine(tmp_path: Path) -> None:
         ),
         {**SECRETS, "NISSE_ENV": str(tmp_path / ".env")},
     )
-    assert config.workspace.name == "project"
+    assert only(config).workspace.name == "project"
 
 
 def test_context_tuning_is_read_and_checked() -> None:
@@ -346,10 +351,10 @@ def test_context_tuning_is_read_and_checked() -> None:
 
 
 def test_the_build_host_is_required_and_defaults_to_root_on_22() -> None:
-    config = build()
-    assert (config.build_host.address, config.build_host.user) == ("build.test", "root")
-    assert config.build_host.port == 22
-    assert config.build_host.key.is_absolute()
+    host = only(build()).build_host
+    assert (host.address, host.user) == ("build.test", "root")
+    assert host.port == 22
+    assert host.key.is_absolute()
     with pytest.raises(ConfigError, match="build_host.address"):
         build(**{f"{P}build_host__address": None})
 
@@ -360,14 +365,14 @@ def test_a_bad_build_host_port_is_reported() -> None:
 
 
 def test_the_push_key_is_optional_and_a_path() -> None:
-    assert build().push_key is None
-    assert build(**{f"{P}push_key": "~/k"}).push_key == Path("~/k").expanduser()
+    assert only(build()).push_key is None
+    assert only(build(**{f"{P}push_key": "~/k"})).push_key == Path("~/k").expanduser()
 
 
 def test_a_push_token_needs_its_user_and_excludes_a_key() -> None:
     token = {f"{P}push_user": "pusher", f"{P}push_token": "${TEST_GITEA_TOKEN}"}
-    config = build(**token)
-    assert (config.push_user, config.push_token) == ("pusher", "gt")
+    project = only(build(**token))
+    assert (project.push_user, project.push_token) == ("pusher", "gt")
     with pytest.raises(ConfigError, match="set both or neither"):
         build(**{f"{P}push_user": "pusher"})
     with pytest.raises(ConfigError, match="not both"):
@@ -386,11 +391,21 @@ def test_no_credential_appears_in_the_config_repr() -> None:
         assert secret not in text
 
 
-def test_the_loop_defaults_to_old_and_takes_new() -> None:
-    assert build().loop == "old"
-    assert build(loop="new").loop == "new"
-    with pytest.raises(ConfigError, match="loop: must be old or new"):
-        build(loop="newer")
+def test_the_old_loops_settings_are_refused_with_what_to_do_instead() -> None:
+    old = tree(loop="new", roles__planner="claude-opus-5", roles__driver=None)
+    old["roles"]["executor"] = "local-fast"
+    old["tuning"] = {"executor": {"max_turns": 40}}
+    with pytest.raises(ConfigError) as excinfo:
+        AgentConfig.from_tree(old, SECRETS)
+    message = str(excinfo.value)
+    for expected in (
+        "loop: retired — the loop is the only one now; remove it",
+        "roles.planner: retired — use roles.driver",
+        "roles.executor: retired",
+        "tuning.executor: retired — it is tuning.driver now",
+        "unset or empty — roles.driver",
+    ):
+        assert expected in message
 
 
 # --- projects (D8) -----------------------------------------------------------
@@ -410,14 +425,6 @@ def test_a_second_project_is_a_second_entry() -> None:
     assert other.base_branch == "main" and other.labels == () and other.secrets == {}
 
 
-def test_the_old_loop_refuses_more_than_one_project() -> None:
-    config = build(projects__other=SECOND)
-    with pytest.raises(ConfigError, match="exactly one project"):
-        config.only_project()
-    with pytest.raises(ConfigError, match="exactly one project"):
-        _ = config.gitea_repo
-
-
 def test_at_least_one_project_is_needed() -> None:
     with pytest.raises(ConfigError, match="projects: must name at least one"):
         build(projects={})
@@ -439,8 +446,8 @@ def test_a_project_name_is_a_slug() -> None:
 
 def test_labels_and_egress_are_lists_of_names() -> None:
     config = build(**{f"{P}labels": ["discuss"], f"{P}egress": ["pypi.org"]})
-    assert config.only_project().labels == ("discuss",)
-    assert config.only_project().egress == ("pypi.org",)
+    assert only(config).labels == ("discuss",)
+    assert only(config).egress == ("pypi.org",)
     with pytest.raises(ConfigError, match="projects.greet.labels: must be a list"):
         build(**{f"{P}labels": "discuss"})
 
@@ -454,7 +461,7 @@ def test_secrets_are_by_reference_with_a_visibility() -> None:
             }
         }
     )
-    project = config.only_project()
+    project = only(config)
     assert project.command_secrets() == {"DB_URL": "sk-x"}
     assert project.secrets["DEPLOY"].visible == "harness"