Roles, capability profiles and help on the new loop (M2.6)

- roles.driver (unset: the planner's model) and roles.classifier (unset:
  the expert's); the mapping is pinned when an item opens, and the driver
  and the go-ahead reader run on the item's own (L9).
- profiles: in configuration (tools, strengths, weaknesses), used in
  briefings: the driver's instructions say what its helper and expert are
  good at, the helper's brief who reads its handback; never a model name.
- delegate(task) and ask_expert(question) for the driver. A helper works
  on a branch off the milestone's; on a handback its work comes back as
  uncommitted changes, otherwise it is discarded, so harness branches
  never hold commits only the clone has. ask_expert keeps its budget.
- Triggers: failure_limit failed commands in a row (Driver's new "stuck"
  outcome) or the turn cap, mid-milestone, call a helper in once per
  milestone; if it fails too, the item waits on the operator. A
  delegation cut off by a crash is discarded at boot.
- Evaluation capture: delegations and ask_expert as before, and the
  reader's verdicts as classifier pairs.

Tests: help tests (merge-back, discard, needs a milestone, the stuck
trigger and its once-per-milestone limit, the expert and its budget,
pinning, briefings by profile, boot discarding a cut-off delegation),
config tests for roles and profiles, and the walk delegating with a
scripted helper that sometimes fails, with a delegations_settle property.
The walk found helper commits stranded on a merged PR's branch, fixed by
returning a helper's work uncommitted.

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

nisse-dev/src/nisse_dev/loop/driver.py

@@ -29,7 +29,8 @@ _PROSE_NUDGE = (
 
 @dataclass(frozen=True)
 class RunOutcome:
-    kind: Literal["wait", "capped", "stopped", "failed"]
+    # stuck: failed commands in a row hit the failure limit (L9's trigger)
+    kind: Literal["wait", "capped", "stuck", "stopped", "failed"]
     tool: str = ""  # the waiting tool called
     arguments: dict[str, Any] | None = None
     note: str = ""
@@ -43,10 +44,12 @@ class Driver:
         *,
         max_turns: int = DEFAULT_MAX_TURNS,
         cancel: Cancel | None = None,
+        failure_limit: int = 0,  # 0: no limit
     ) -> None:
         self._client = client
         self._model = model
         self._max_turns = max_turns
+        self._failure_limit = failure_limit
         self._cancel = cancel or Cancel()
 
     def run(
@@ -59,6 +62,7 @@ class Driver:
         handlers: Mapping[str, Callable[[dict[str, Any]], tuple[str, bool]]]
         | None = None,
         checks: Mapping[str, Callable[[dict[str, Any]], str | None]] | None = None,
+        model: str | None = None,
     ) -> RunOutcome:
         """``handlers`` answer control tools that don't end the run (e.g.
         start_milestone): each returns a result and whether it succeeded.
@@ -66,6 +70,8 @@ class Driver:
         failed result and the run goes on, so the model puts it right."""
         handlers = handlers or {}
         checks = checks or {}
+        model = model or self._model  # the work item's pinned driver (L9)
+        failures: list[str] = []  # the failed commands in a row, latest last
         offered = names(tools)
         nudged = False
         for turn in range(self._max_turns):
@@ -74,7 +80,7 @@ class Driver:
                 return RunOutcome("stopped", note=f"before turn {turn + 1}")
             try:
                 produced = self._client.complete(
-                    transcript.view(), tools=tools, model=self._model, system=system
+                    transcript.view(), tools=tools, model=model, system=system
                 )
             except LLMError as exc:
                 return RunOutcome("failed", note=f"the driver's call failed: {exc}")
@@ -132,4 +138,21 @@ class Driver:
                 transcript.append(
                     ToolResult(call_id=call.call_id, result=result, success=ok)
                 )
+                if call.tool == "run_command":
+                    failures = [] if ok else [*failures, _brief(call, result)]
+                    if self._failure_limit and len(failures) >= self._failure_limit:
+                        skip_calls(
+                            transcript, calls[index + 1 :], "not run: you're stuck"
+                        )
+                        return RunOutcome(
+                            "stuck",
+                            note=f"{len(failures)} commands failed in a row: "
+                            + "; ".join(failures),
+                        )
         return RunOutcome("capped", note=f"hit the {self._max_turns}-turn cap")
+
+
+def _brief(call: ToolCall, result: str) -> str:
+    """A failed command and the end of what it said, for a helper's brief."""
+    command = str(call.arguments.get("command", ""))[:120]
+    return f"`{command}`: {result.strip()[-300:]}"