Merge pull request 'Fixes from the first live session of the new loop' (#119) from live-fixes into main

Found live on 2026-10-07 (bring-up on nisse-testing-operator/nisse-bringup,
Opus as driver):

- The driver's prose nudge went into the system prompt, so the retry ended
  on the model's own reply. Opus 5.5 refuses that ("does not support
  assistant message prefill. The conversation must end with a user
  message"). The nudge is now a harness fact in the conversation. The
  delegation harness's prose path did the same and is fixed likewise.
- A helper's first request ended with its own delegation_started event,
  rendered as an assistant message: every delegation would have failed on
  Opus. The helper now sees it as the harness handing over its task.
- model_contract checks that every request ends on the user's side, so
  scripted tests catch this from now on (each fix is mutation-checked).
- `status` (and `details`) in an item's thread were taken as replies: they
  woke the driver and cancelled a bug's countdown. They now answer at once
  and change nothing.
- The driver had no way to answer a remark without changing the item, so it
  answered in prose. Every run woken by a message (or a review comment) now
  offers `reply`, which answers where asked and returns to the very wait
  that was interrupted.

Also: scripts/bringup-repo.sh makes the bring-up's test repository
(make_repo/delete_repo, split out of fresh_repo), and made repositories get
the agent's labels (bug, discuss, design).

Tests: test_loop_live_fixes.py (one per finding); the walk sends `status`
into threads (status_changes_nothing) and its scripted driver sometimes
answers remarks with reply.

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

.gitea/workflows/ci.yml

@@ -1,4 +1,10 @@
 name: CI
+# Three levels of testing:
+#   - every PR: lint, unit tests (16 property walks), integration tests, and
+#     the smoke end-to-end tests (`smoke` marker);
+#   - a PR labelled e2e:full, a push to main and a manual run: all 48 walks
+#     and the whole end-to-end suite;
+#   - a published release: the full set again before its image (release.yml).
 on:
   # Only main on push. A PR branch is already covered by the pull_request
   # event, and running both fired two identical suites concurrently — which
@@ -7,11 +13,18 @@ on:
   push:
     branches: [main]
   pull_request:
+    # labeled: adding e2e:full runs the full set; any other label, nothing.
+    types: [opened, synchronize, reopened, labeled]
   # Manual re-run, e.g. after rotating a credential or raising a budget cap.
   workflow_dispatch:
 
+env:
+  # Full on e2e:full, and outside PRs.
+  FULL: ${{ github.event_name != 'pull_request' || contains(github.event.pull_request.labels.*.name, 'e2e:full') }}
+
 jobs:
   checks:
+    if: github.event.action != 'labeled' || github.event.label.name == 'e2e:full'
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
@@ -26,12 +39,10 @@ jobs:
 
       - name: Install sshd
         # The tests reach "the build VM" over real SSH, through a throwaway
-        # sshd the test fixture starts (nisse_dev/testing.py).
+        # sshd the test fixture starts (nisse_dev/testing.py). Skipped when the image has it; otherwise from
-        run: |
+        # a kernel.org mirror: archive.ubuntu.com has been slow enough to add
-          SUDO=$(command -v sudo || true)
+        # minutes to every job.
-          $SUDO apt-get update -qq
+        run: scripts/ci-sshd.sh
-          $SUDO apt-get install -y -qq openssh-server openssh-client > /dev/null
-          $SUDO mkdir -p /run/sshd
 
       - name: Lint
         run: uv run ruff check .
@@ -40,7 +51,12 @@ jobs:
         run: uv run ruff format --check .
 
       - name: Unit tests
-        run: uv run pytest -m "not integration and not e2e"
+        # Across the runner's cores (pytest-xdist). On a PR, 16 property
+        # walks from a seed that moves with the run number; all 48 otherwise.
+        env:
+          NISSE_WALK_SEEDS: ${{ env.FULL == 'true' && '48' || '16' }}
+          NISSE_WALK_FIRST: ${{ env.FULL == 'true' && '0' || github.run_number }}
+        run: uv run pytest -n auto -m "not integration and not e2e"
 
       - name: Integration tests
         # Everything but the secrets is in tests/integration/nisse-test.yaml,
@@ -58,12 +74,15 @@ jobs:
   e2e:
     # The end-to-end suite (scripts/e2e.sh): the whole agent against real
     # services, every role on a local model, so it spends nothing but keeps
-    # the RTX 4090 busy for ~25 minutes. On PRs only, and only after the
+    # the RTX 4090 busy: ~10 minutes for the smoke tests, ~25+ for all. On
-    # checks pass: a PR that's red already doesn't spend the GPU. There is
+    # PRs and manual runs, and only after the checks pass: a PR that's red
-    # one runner and Gitea doesn't honour cancel-in-progress, so a run a
+    # already doesn't spend the GPU. There is one runner and Gitea doesn't
-    # newer push supersedes holds the runner until it's cancelled by hand.
+    # honour cancel-in-progress, so a run a newer push supersedes holds the
-    # Locally, run only what you're writing (-k) or diagnosing (--lf).
+    # runner until it's cancelled by hand. Locally, run only what you're
-    if: github.event_name != 'push'
+    # writing (-k) or diagnosing (--lf).
+    if: >-
+      github.event_name != 'push'
+      && (github.event.action != 'labeled' || github.event.label.name == 'e2e:full')
     needs: checks
     runs-on: ubuntu-latest
     timeout-minutes: 90
@@ -79,12 +98,10 @@ jobs:
         run: uv sync --all-packages
 
       - name: Install sshd
-        # The build VM is a throwaway sshd here (scripts/e2e.sh --local).
+        # The build VM is a throwaway sshd here (scripts/e2e.sh --local). Skipped when the image has it; otherwise from
-        run: |
+        # a kernel.org mirror: archive.ubuntu.com has been slow enough to add
-          SUDO=$(command -v sudo || true)
+        # minutes to every job.
-          $SUDO apt-get update -qq
+        run: scripts/ci-sshd.sh
-          $SUDO apt-get install -y -qq openssh-server openssh-client > /dev/null
-          $SUDO mkdir -p /run/sshd
 
       - name: End-to-end tests
         env:
@@ -93,4 +110,9 @@ jobs:
           NISSE_GITEA_TOKEN: ${{ secrets.NISSE_GITEA_TOKEN }}
           NISSE_OPERATOR_TOKEN: ${{ secrets.NISSE_OPERATOR_TOKEN }}
           NISSE_E2E_RUNS: /tmp/nisse-e2e-runs
-        run: scripts/e2e.sh --local -v
+        run: |
+          if [ "$FULL" = true ]; then
+            scripts/e2e.sh --local -v
+          else
+            scripts/e2e.sh --local -v -m "e2e and smoke"
+          fi

.gitea/workflows/release.yml

@@ -3,7 +3,8 @@ name: Release
 # registry (Development Plan M1.5, X4): gitea.dongar.net/dev/nisse-dev
 # tagged with the version (the release tag without its v) and latest. Cut a
 # release in Gitea (Releases → New release, tag vX.Y.Z); a tag pushed for any
-# other reason builds nothing.
+# other reason builds nothing. The full test set runs first: no image without
+# it passing.
 #
 # The registry needs a personal access token with write:package
 # (REGISTRY_TOKEN, an organization secret of `dev`, belonging to the
@@ -23,7 +24,46 @@ env:
   IMAGE: gitea.dongar.net/dev/nisse-dev
 
 jobs:
+  tests:
+    # Everything, before an image exists: lint, unit tests with all 48
+    # property walks, integration tests and the whole end-to-end suite.
+    runs-on: ubuntu-latest
+    timeout-minutes: 150
+    env:
+      NISSE_LLM_TOKEN: ${{ secrets.NISSE_LLM_TOKEN }}
+      NISSE_MATRIX_TOKEN: ${{ secrets.NISSE_MATRIX_TOKEN }}
+      NISSE_GITEA_TOKEN: ${{ secrets.NISSE_GITEA_TOKEN }}
+      NISSE_OPERATOR_TOKEN: ${{ secrets.NISSE_OPERATOR_TOKEN }}
+      NISSE_E2E_RUNS: /tmp/nisse-e2e-runs
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Install uv
+        run: |
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          echo "$HOME/.local/bin" >> "$GITHUB_PATH"
+
+      - name: Sync workspace
+        run: uv sync --all-packages
+
+      - name: Install sshd
+        run: scripts/ci-sshd.sh
+
+      - name: Lint
+        run: uv run ruff check . && uv run ruff format --check .
+
+      - name: Unit tests
+        run: uv run pytest -n auto -m "not integration and not e2e"
+
+      - name: Integration tests
+        run: uv run pytest -m "integration and not e2e" -rs
+
+      - name: End-to-end tests
+        run: scripts/e2e.sh --local -v
+
   build-and-push:
+    needs: tests
     runs-on: ubuntu-latest
     permissions:
       contents: read

Dockerfile

@@ -10,6 +10,12 @@
 #   docker build -t nisse-dev .                  the agent
 #   docker build --target test -t nisse-dev:test .   plus the tests, for
 #                                                   scripts/e2e.sh --image
+#   docker build --target test --build-arg TEST_BASE=<published image> \
+#       -t nisse-dev:test .                      the tests over a release as
+#                                                published (M1.5 acceptance)
+
+# What the test stage stacks on: this build's agent, or a published image.
+ARG TEST_BASE=runtime
 
 FROM python:3.12-slim AS build
 COPY --from=ghcr.io/astral-sh/uv:0.12.10 /uv /usr/local/bin/uv
@@ -40,9 +46,10 @@ VOLUME /var/lib/nisse-dev
 USER nisse
 ENTRYPOINT ["nisse-dev"]
 
-# The image's own package plus the tests, so the end-to-end suite runs
+# The tests over an image's own package (this build's, or a published one,
-# against what was built, not the source tree.
+# TEST_BASE): only the test tools are added, never the agent reinstalled, so
-FROM runtime AS test
+# the end-to-end suite runs against what was built, not the source beside it.
+FROM ${TEST_BASE} AS test
 USER root
 RUN apt-get update \
     && apt-get install -y --no-install-recommends git openssh-server \
@@ -57,11 +64,9 @@ COPY nisse-ops nisse-ops
 COPY tests tests
 COPY scripts scripts
 COPY conftest.py ./
-ENV UV_PROJECT_ENVIRONMENT=/opt/nisse
+ENV UV_PROJECT_ENVIRONMENT=/opt/nisse UV_NO_SYNC=1
-# Test tools too, all installed (not editable), and `uv run` never re-syncs:
+RUN uv export --frozen --only-group dev --no-hashes --no-emit-project > /tmp/dev.txt \
-# the tests import the package as built, not the source tree beside them.
+    && uv pip install --quiet --python /opt/nisse/bin/python -r /tmp/dev.txt
-RUN uv sync --frozen --all-packages --no-editable --inexact
-ENV UV_NO_SYNC=1
 ENTRYPOINT []
 
 # The default target: the agent itself.

docs/Development-Plan.md

@@ -73,10 +73,14 @@ Only `nisse-dev` has real code; `nisse-core` and `nisse-ops` are placeholders.
   `CLAUDE.md`. A milestone marked **needs discussion** starts with a design
   discussion first.
 - **Every change comes with tests,** through `scripts/check.sh`.
-- **The end-to-end test runs on every PR** (M1.6): the real agent, every role on a
+- **Testing has three levels** (M1.6): every PR runs the checks (unit tests across
-  local model (so it costs nothing per call), a repository of the run's own and a
+  the runner's cores, 16 property walks from a moving seed, integration) and the
-  real build host, driven by a scripted operator. CI runs it as its own job on pull
+  end-to-end **smoke** tests; a PR labelled `e2e:full`, a push to `main` and a
-  requests, after the checks pass. There is one runner, and a run a newer push
+  manual run add all 48 walks and the whole end-to-end suite; a published release
+  runs that full set before its image is built. The end-to-end tests drive the
+  real agent, every role on a local model (so they cost nothing per call), in a
+  repository of the run's own, with a scripted operator; CI runs them after the
+  checks pass. There is one runner, and a run a newer push
   supersedes has to be cancelled by hand. Locally only the tests being written
   (`-k`) or diagnosed (`--lf`); `scripts/e2e.sh` reports what a run spent. Manual
   live runs are for the new loop's first runs and phase ends. Findings are filed as
@@ -271,8 +275,9 @@ in prose, asking mid-work (L5, #42, #77).
 - Deferred: config reload without a restart (with M2.5's entry step if wanted);
   the trust dial's configuration (M3.3).
 
-**M1.5 The agent as a container image.** *Pre-approved.* **Built** (this PR);
+**M1.5 The agent as a container image.** *Pre-approved.* **Done** (#115; accepted
-accepted once a tag's image passes the e2e suite.
+2026-10-07: `v0.1.0`'s published image passed the whole e2e suite, 9 of 9, run with
+the release's own tests over it).
 - A Dockerfile for `nisse-dev`, stateless with a volume for state; a Gitea Actions
   workflow that builds it and pushes to Gitea's registry on each release tag (X4).
 - `bringup.sh` runs the image instead of a venv.
@@ -283,6 +288,10 @@ accepted once a tag's image passes the e2e suite.
   package, for `scripts/e2e.sh --image REF`. Images are built only on release:
   `release.yml` pushes `gitea.dongar.net/dev/nisse-dev:<version>` and `:latest`
   on a published release (`REGISTRY_TOKEN`, the `dev` organization's).
+- Acceptance runs the tests over a published image: `docker build --target test
+  --build-arg TEST_BASE=gitea.dongar.net/dev/nisse-dev:<version>` from the
+  release's checkout, then `scripts/e2e.sh --image`; the test stage adds only the
+  test tools, so the agent under test is the one published.
 - The live bring-up's switch to the image waits for the next live test (it needs
   its own test repository too); `bringup.sh plan` shows what it will run.
 

nisse-dev/src/nisse_dev/delegation.py

@@ -33,6 +33,7 @@ from nisse_dev.canonical import (
     AssistantText,
     DelegationCompleted,
     DelegationStarted,
+    Fact,
     PlanSubmitted,
     StoredEvent,
     ToolCall,
@@ -414,6 +415,15 @@ class Delegation:
                 # and its tree state is unknown. Treat it as a failure rather
                 # than inventing a handback on its behalf.
                 if any(isinstance(e, AssistantText) for e in produced):
+                    # Said something but did nothing: told so, as the
+                    # harness, so the next request ends on the user's side
+                    # (models refuse assistant prefill).
+                    transcript.as_actor("orchestrator").append(
+                        Fact(
+                            text="Your last reply was prose. Use your tools to "
+                            "do the work, then call hand_back."
+                        )
+                    )
                     continue
                 return DelegationResult(
                     "failed",

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

@@ -116,6 +116,10 @@ asking is better than guessing.
 Before each group of tool calls, say in a sentence what you're doing and why:
 the operator sees it as your progress.
 
+When the operator says something that needs an answer but no change (a
+remark, a status question), answer with reply: the item goes back to what it
+was waiting for. Every reply of yours is a tool call, never bare prose.
+
 The project's documents are in its repository: read ARCHITECTURE.md and any
 build plans under docs/plans/ before you restate.
 
@@ -587,6 +591,8 @@ class LoopAgent:
                 loop, position = replace(loop, pr=0, pr_url="", branch=""), READY
             elif new and item.position == WAITING:
                 position = READY
+                # A comment can be answered with a reply, back to this wait.
+                loop = replace(loop, back_to=loop.wait, back_note=loop.note)
             if position != item.position or loop != loop_of(item):
                 if position == READY:
                     loop = replace(loop, wait="", note="")
@@ -824,6 +830,19 @@ class LoopAgent:
             elif not self._store.active_item():
                 self._transport.post("Nothing is running.", thread=msg.thread)
             return
+        if command in ("status", "details"):
+            # About the item, not a reply to it: nothing changes, a
+            # countdown runs on (L5). Recorded at the agent's level.
+            self._store.transcript(None, "operator").append(
+                UserMessage(text=text, source=f"transport:{msg.sender}")
+            )
+            said = (
+                self._details(item)
+                if command == "details"
+                else self._status_line(item, [i.id for i in self._queue()])
+            )
+            self._transport.post(said, thread=loop_of(item).thread)
+            return
         self._store.transcript(item.id, "operator").append(
             UserMessage(text=text, source=f"transport:{msg.sender}")
         )
@@ -844,8 +863,6 @@ class LoopAgent:
             self._close(item, loop, "abandoned", f"abandoned by {msg.sender}")
         elif command == "done" and loop.new_project:
             self._provisioned(item, loop)
-        elif command == "details":
-            self._transport.post(self._details(item), thread=loop.thread)
         elif (
             item.position == WAITING
             and item.scope != GENERAL
@@ -903,12 +920,13 @@ class LoopAgent:
             )
             self._go(item, loop, f"{msg.sender} (in plain language)")
         elif item.position == WAITING:
-            # An answer, or changes to a restatement: the driver reads it.
+            # An answer, changes to a restatement, or a remark: the driver
+            # reads it, and may reply and go back to this wait (L5).
             save(
                 self._store,
                 item.id,
                 READY,
-                replace(loop, wait="", note=""),
+                replace(loop, wait="", note="", back_to=loop.wait, back_note=loop.note),
                 cause="the operator answered",
             )
         self._schedule()
@@ -1272,8 +1290,9 @@ class LoopAgent:
                 f"{self._repos()}."
             )
         queue = [i.id for i in self._queue()]
-        lines = []
+        return "\n".join(self._status_line(item, queue) for item in items)
-        for item in items:
+
+    def _status_line(self, item: WorkItem, queue: list[str]) -> str:
         loop = loop_of(item)
         what = item.position
         if item.position == WAITING and loop.wait == CHAT:
@@ -1292,8 +1311,7 @@ class LoopAgent:
             if item.scope == GENERAL
             else f"{self._label(item)} {item.title}"
         )
-            lines.append(f"{name}: {what}")
+        return f"{name}: {what}"
-        return "\n".join(lines)
 
     # -- running (L1, L2) ------------------------------------------------------
 
@@ -1663,12 +1681,15 @@ class LoopAgent:
                     CLOSE_ITEM,
                     *([WAIT_FOR_REVIEW] if loop.pr else []),
                     *([DELEGATE, ASK_EXPERT] if self._helpers else []),
+                    *([REPLY] if loop.back_to else []),
                     *EXECUTOR_TOOLS,
                 ]
             ),
         ]
         if loop.question:
             tools = [REPLY, *_LOOK]
+        elif loop.back_to and not loop.started:
+            tools = [*tools, REPLY]  # before the go-ahead, a remark answered
         roles = self._roles_of(item)
         workspace = project.workspace
         system = SYSTEM.format(
@@ -1771,6 +1792,13 @@ class LoopAgent:
                 replace(loop, rev=rev, wait=GO_AHEAD, note="", grace_until=grace),
                 None,
             )
+        elif outcome.kind == "wait" and outcome.tool == "reply" and loop.back_to:
+            # A project item's reply: answered where asked (the thread or the
+            # PR), then back to the very wait the message interrupted (L5).
+            self._answer_where_asked(item, loop, str(args.get("text", "")).strip())
+            self._wait(
+                item, replace(loop, wait=loop.back_to, note=loop.back_note), None
+            )
         elif outcome.kind == "wait" and outcome.tool == "reply":
             text = str(args.get("text", "")).strip()
             self._transport.post(text, thread=loop.thread)
@@ -2068,6 +2096,8 @@ class LoopAgent:
     def _wait(self, item: WorkItem, loop: Loop, say: str | None) -> None:
         if say:
             self._transport.post(say, thread=loop.thread)
+        # Waiting again: whatever woke the item is settled.
+        loop = replace(loop, question=False, back_to="", back_note="")
         save(self._store, item.id, WAITING, loop, cause=f"waiting for {loop.wait}")
 
     def _close(self, item: WorkItem, loop: Loop, outcome: str, note: str) -> None:

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

@@ -12,7 +12,7 @@ from dataclasses import dataclass
 from typing import Any, Literal
 
 from nisse_dev.cancel import Cancel, skip_calls
-from nisse_dev.canonical import AssistantText, ToolCall, ToolResult
+from nisse_dev.canonical import AssistantText, Fact, ToolCall, ToolResult
 from nisse_dev.llm import LLMClient, LLMError
 from nisse_dev.loop.tools import WAITING, names
 from nisse_dev.state import Transcript
@@ -100,7 +100,10 @@ class Driver:
                 if nudged:
                     return RunOutcome("failed", note="kept replying in prose")
                 nudged = True
-                system = f"{system}\n\n{_PROSE_NUDGE}"
+                # In the conversation, as the harness: a request must end on
+                # the user's side (models refuse assistant prefill), and the
+                # nudge is then in the log as what it was.
+                transcript.as_actor("orchestrator").append(Fact(text=_PROSE_NUDGE))
                 continue
             for index, call in enumerate(calls):
                 if self._cancel.requested:

nisse-dev/src/nisse_dev/views.py

@@ -22,6 +22,7 @@ from dataclasses import replace
 from nisse_dev.canonical import (
     AssistantText,
     DelegationCompleted,
+    DelegationStarted,
     Fact,
     GateOpened,
     PlanSubmitted,
@@ -105,6 +106,13 @@ def project(events: list[StoredEvent], viewer: str) -> list[StoredEvent]:
             if viewer == "helper":
                 out.append(_said(stored, "approved plan", event.plan))
             continue
+        if isinstance(event, DelegationStarted):
+            # The harness handing a task over: to the helper given it, a
+            # message from the harness, never its own turn (a request must
+            # end on the user's side; models refuse assistant prefill).
+            if actor == viewer:
+                out.append(_said(stored, "orchestrator", f"your task: {event.brief}"))
+            continue
         if actor == viewer:
             out.append(stored)
         elif actor == "operator" and isinstance(event, UserMessage):

nisse-dev/tests/model_contract.py

@@ -20,6 +20,13 @@ def assert_renderable(events: list[Any]) -> None:
 
 
 def _anthropic(messages: list[dict[str, Any]]) -> None:
+    # A request ends on the user's side: current models refuse assistant
+    # prefill ("The conversation must end with a user message"), found live
+    # on 2026-10-07 when the driver's prose nudge left its own reply last.
+    assert not messages or messages[-1]["role"] == "user", (
+        f"anthropic messages.{len(messages) - 1}: the conversation ends with an "
+        "assistant message (prefill), which the API refuses"
+    )
     for index, message in enumerate(messages):
         if message["role"] != "assistant":
             continue

nisse-dev/tests/test_loop_live_fixes.py

@@ -0,0 +1,154 @@
+"""What the first live session of the new loop found (2026-10-07), each as a
+test: a request ending on the assistant's side (refused by models without
+prefill: the driver's prose nudge, a helper's first request), `status` in an
+item's thread taken as a reply, and no way for the driver to answer a remark.
+"""
+
+from pathlib import Path
+from typing import Any
+
+from model_contract import assert_renderable
+from test_loop import RESTATED, FakeGitea, Harness, ScriptedModel, call, issue
+from test_loop_help import ROLES, hand_back
+from test_loop_milestones import RESTATE, Forge, loop_agent, say
+from test_loop_milestones import item as the_item
+from test_loop_milestones import world as world  # noqa: PLC0414 -- the fixture
+
+from nisse_dev.canonical import AssistantText, Fact
+from nisse_dev.delegation import Caps
+from nisse_dev.gitea import Comment
+from nisse_dev.loop.help import Helpers
+from nisse_dev.loop.items import GO_AHEAD, MERGE, WAITING, loop_of
+
+
+class Clock:
+    def __init__(self) -> None:
+        self.now = 1000.0
+
+    def __call__(self) -> float:
+        return self.now
+
+
+def test_the_prose_nudge_keeps_requests_ending_on_the_user_side(
+    tmp_path: Path,
+) -> None:
+    # ScriptedModel checks every request as the API would, now including
+    # "the conversation must end with a user message".
+    h = Harness(
+        tmp_path,
+        ScriptedModel([AssistantText(text="Let me think.", model="m")], RESTATED),
+        FakeGitea(issue()),
+    )
+    h.agent.intake()
+    assert loop_of(h.item()).wait == GO_AHEAD  # the nudge worked
+    facts = [
+        e.event.text
+        for e in h.store.transcript(h.item().id, "orchestrator").read()
+        if isinstance(e.event, Fact)
+    ]
+    assert any("Your last reply was prose" in f for f in facts)
+
+
+class ContractHelper:
+    """A helper that checks its requests as the API would."""
+
+    def __init__(self, *turns: list[Any]) -> None:
+        self.turns = list(turns)
+
+    def complete(self, events, tools, model, system=None) -> list[Any]:
+        assert_renderable(events)
+        return self.turns.pop(0)
+
+
+def test_a_helpers_requests_end_on_the_user_side(world: dict[str, Any]) -> None:
+    helper = ContractHelper(
+        [AssistantText(text="Looking around first.", model="h")],  # prose
+        [call("write_file", path="helped.py", content="H = 1\n")],
+        [hand_back()],
+    )
+    model = ScriptedModel(
+        RESTATE,
+        [
+            call("start_milestone", name="words"),
+            call("delegate", task="write helped.py"),
+            call("finish_milestone", summary="with help", final="yes"),
+        ],
+    )
+    helpers = Helpers(
+        client=helper,  # type: ignore[arg-type]
+        roles=ROLES,
+        client_for=lambda model: helper,  # type: ignore[arg-type,return-value]
+        caps=Caps(turns=5),
+    )
+    agent, transport = loop_agent(world, model, helpers=helpers)
+    agent.intake()
+    say(transport, world, "go")
+    forge: Forge = world["forge"]
+    [pr] = forge.prs
+    assert forge.files(pr) == ["helped.py"]
+
+
+def test_status_in_a_thread_is_not_a_reply(tmp_path: Path) -> None:
+    clock = Clock()
+    model = ScriptedModel(RESTATED)  # called once only: status runs nothing
+    h = Harness(
+        tmp_path,
+        model,
+        FakeGitea(issue(labels=("bug",))),
+        clock=clock,
+        bug_grace_s=600,
+    )
+    h.agent.intake()
+    grace = loop_of(h.item()).grace_until
+    h.say("status")
+    assert h.posted()[-1].startswith("#14 Add count_words: waiting for go-ahead")
+    loop = loop_of(h.item())
+    assert (h.item().position, loop.wait, loop.grace_until) == (
+        WAITING,
+        GO_AHEAD,
+        grace,
+    )
+    h.say("details")
+    assert loop_of(h.item()).grace_until == grace
+
+
+def test_a_remark_gets_a_reply_and_the_restatement_stays_up(tmp_path: Path) -> None:
+    h = Harness(
+        tmp_path,
+        ScriptedModel(
+            RESTATED,
+            [call("reply", text="Noted: I'll keep it to text.py.")],
+        ),
+        FakeGitea(issue()),
+    )
+    h.agent.intake()
+    h.say("just a thought: keep it small")
+    loop = loop_of(h.item())
+    assert (loop.wait, loop.rev) == (GO_AHEAD, 1)
+    assert h.posted()[-1] == "Noted: I'll keep it to text.py."
+
+
+def test_a_review_comment_can_be_answered_without_a_fix(
+    world: dict[str, Any],
+) -> None:
+    forge: Forge = world["forge"]
+    model = ScriptedModel(
+        RESTATE,
+        [
+            call("start_milestone", name="words"),
+            call("write_file", path="w.py", content="W = 1\n"),
+            call("finish_milestone", summary="words", final="yes"),
+        ],
+        [call("reply", text="It's a constant on purpose, for the test.")],
+    )
+    agent, transport = loop_agent(world, model)
+    agent.intake()
+    say(transport, world, "go")
+    [pr] = forge.prs
+    forge.comments[pr] = [Comment(7, "don", "why a constant", "t")]
+    agent.watch()
+    agent._schedule()
+    loop = loop_of(the_item(world))
+    assert (loop.wait, loop.pr) == (MERGE, pr)
+    assert forge.posted[-1][0] == pr  # answered on the PR, where asked
+    assert "It's a constant on purpose" in forge.posted[-1][1]

nisse-dev/tests/test_loop_properties.py

@@ -8,6 +8,7 @@ checks every property after every step. A failure prints
 the seed and the steps, so it replays exactly.
 """
 
+import os
 import random
 import re
 import subprocess
@@ -46,8 +47,14 @@ from nisse_dev.tools import Workspace
 from nisse_dev.transport import InboundMessage, InMemoryTransport
 from nisse_dev.worker import InlineWorker
 
-SEEDS = range(32)
+# How many walks, and from which seed: all 48 by default; CI runs fewer on a
-TWO_PROJECT_SEEDS = range(100, 116)
+# PR, from a seed that moves with the run number, so PRs still cover new
+# ground (NISSE_WALK_SEEDS, NISSE_WALK_FIRST). Two thirds walk one project,
+# a third two.
+_WALKS = int(os.environ.get("NISSE_WALK_SEEDS") or 48)
+_FIRST = int(os.environ.get("NISSE_WALK_FIRST") or 0)
+SEEDS = range(_FIRST, _FIRST + _WALKS * 2 // 3)
+TWO_PROJECT_SEEDS = range(100 + _FIRST, 100 + _FIRST + _WALKS - _WALKS * 2 // 3)
 STEPS = 30
 
 
@@ -76,8 +83,15 @@ class PolicyModel:
             return [call("ask_human", question=f"question {self.n}?")]
         if roll < 0.12:
             return [ToolCall(call_id=f"p{self.n}", tool="nonsense", arguments={})]
-        if "reply" in offered and "propose_work" not in offered:
+        if "reply" in offered and not offered & {
+            "propose_work",
+            "restate",
+            "ask_human",
+            "start_milestone",
+        }:
             return [call("reply", text=f"because {self.n}")]  # a question run
+        if "reply" in offered and "propose_work" not in offered and roll < 0.2:
+            return [call("reply", text=f"noted {self.n}")]  # a remark answered
         if "reply" in offered:  # a conversation: talk, or propose work
             if self.rng.random() < 0.1:
                 return [
@@ -251,6 +265,7 @@ class Walk:
         if item and loop:
             weights.update({"go": 3, "answer": 2, "abandon": 1, "go-stale": 1})
             weights.update({"plain-yes": 2, "wait": 2, "question": 2})
+            weights.update({"status-here": 1})
             if loop.new_project:
                 weights.update({"done": 2})
             if item.scope != "general":  # conversations aren't named by number
@@ -266,6 +281,18 @@ class Walk:
         elif action == "done" and loop:
             # The operator says a new project is ready; no entry was added.
             self.transport.receive(DON, "done", thread=loop.thread)
+        elif action == "status-here" and loop:
+            # L5: a command about the item, not a reply: nothing changes,
+            # not even a countdown.
+            before = (item.position, loop.wait, loop.rev, loop.grace_until)
+            self.transport.receive(DON, "status", thread=loop.thread)
+            after_item = self.world["store"].work_item(item.id)
+            after = loop_of(after_item)
+            now = (after_item.position, after.wait, after.rev, after.grace_until)
+            assert now == before, (
+                f"status_changes_nothing: {before} -> {now}: "
+                f"seed {self.seed}, steps {self.steps}"
+            )
         elif action == "question" and loop:
             # L5: a question never changes the item it's asked about.
             before = (item.position, loop.wait, loop.rev, loop.pr, loop.branch)

pyproject.toml

@@ -4,6 +4,7 @@ members = ["nisse-core", "nisse-dev", "nisse-ops"]
 [dependency-groups]
 dev = [
     "pytest>=8",
+    "pytest-xdist>=3",
     "ruff>=0.6",
 ]
 
@@ -19,6 +20,7 @@ addopts = "--strict-markers --ignore=tests/integration/seed"
 markers = [
     "integration: end-to-end tests that exercise installed entry points or external services",
     "e2e: the whole agent against real services, run after each milestone (scripts/e2e.sh)",
+    "smoke: the end-to-end tests every PR runs; the rest on e2e:full and before a release",
 ]
 
 [tool.ruff]

scripts/bringup-repo.sh

@@ -0,0 +1,42 @@
+#!/usr/bin/env bash
+# The bring-up's test repository (Development Plan, live runs):
+# nisse-testing-operator/nisse-bringup, made as the end-to-end tests make
+# theirs (tests/integration/e2e_support.py, make_repo), but kept between runs.
+# Seeded with the test package, the bring-up agent's account a collaborator,
+# main protected, and a deploy key of its own at $E2E/ssh/nisse-bringup.key.
+#
+#   scripts/bringup-repo.sh create    make it
+#   scripts/bringup-repo.sh reset     delete it and make it again
+#   scripts/bringup-repo.sh delete    delete it
+#
+# Needs NISSE_OPERATOR_TOKEN (.env here) and the bring-up agent's token
+# (NISSE_GITEA_TOKEN in $E2E/.env), which names the agent's account.
+set -euo pipefail
+cd "$(dirname "$0")/.."
+E2E=${NISSE_E2E:-/root/nisse-e2e}
+case "${1:-}" in create|reset|delete) ;; *) sed -n '2,15p' "$0" | sed 's/^# \{0,1\}//'; exit 2 ;; esac
+set -a; [ -f .env ] && . ./.env; set +a
+exec uv run python - "$1" "$E2E" <<'PY'
+import sys
+from pathlib import Path
+
+sys.path.insert(0, "tests/integration")
+from e2e_support import OWNER, delete_repo, make_repo, operator_gitea  # noqa: E402
+
+from nisse_dev.envfile import load_env_file  # noqa: E402
+from nisse_dev.gitea import GiteaClient  # noqa: E402
+
+action, e2e = sys.argv[1], Path(sys.argv[2])
+url = "https://gitea.dongar.net"
+operator = operator_gitea(url)
+name = "nisse-bringup"
+if action in ("reset", "delete"):
+    delete_repo(operator, f"{OWNER}/{name}")
+    print(f"deleted {OWNER}/{name}")
+if action in ("create", "reset"):
+    agent_env: dict[str, str] = {}
+    load_env_file(e2e / ".env", agent_env)
+    agent = GiteaClient(url, agent_env["NISSE_GITEA_TOKEN"]).whoami()
+    repo = make_repo(operator, name, e2e / "ssh", agent=agent)
+    print(f"made {repo.name}: {agent} a collaborator; push key {repo.push_key}")
+PY

scripts/ci-sshd.sh

@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+# The throwaway sshd the tests need (nisse_dev/testing.py), in a CI job.
+# Skipped when the job's image already has it. Otherwise apt fetches from a
+# kernel.org mirror rather than archive.ubuntu.com, which has been slow
+# enough (a few KB/s) to add minutes to every job.
+set -euo pipefail
+SUDO=$(command -v sudo || true)
+if command -v /usr/sbin/sshd > /dev/null && command -v ssh > /dev/null; then
+    echo "sshd already installed"
+else
+    MIRROR=${APT_MIRROR:-http://mirrors.edge.kernel.org/ubuntu}
+    # Both layouts: deb822 (ubuntu.sources, 24.04+) and sources.list (older).
+    for f in /etc/apt/sources.list.d/ubuntu.sources /etc/apt/sources.list; do
+        [ -f "$f" ] && $SUDO sed -i -E \
+            "s#https?://(archive|security)\.ubuntu\.com/ubuntu/?#$MIRROR/#g" "$f"
+    done
+    $SUDO apt-get update -qq
+    $SUDO apt-get install -y -qq openssh-server openssh-client > /dev/null
+fi
+$SUDO mkdir -p /run/sshd

scripts/e2e.sh

@@ -23,6 +23,8 @@ if [ "${1:-}" = "--image" ]; then
     mkdir -p "$runs"
     secrets=()
     [ -f .env ] && secrets=(--env-file .env)
+    # The model, if overridden here, reaches the tests inside too.
+    [ -n "${NISSE_E2E_MODEL:-}" ] && secrets+=(-e "NISSE_E2E_MODEL=$NISSE_E2E_MODEL")
     # Host networking: the suite reaches Matrix, Gitea and LiteLLM as here.
     exec docker run --rm --network host "${secrets[@]}" \
         -e NISSE_E2E_RUNS=/runs -v "$runs:/runs" \

tests/integration/e2e_support.py

@@ -15,8 +15,10 @@ from nisse_dev.llm import RoleMap
 
 # The end-to-end tests run every role on a local model, which costs nothing
 # per call; NISSE_E2E_MODEL overrides it (e.g. claude-haiku-4-5 for a quick,
-# paid run).
+# paid run). The no-thinking alias (LiteLLM: chat_template_kwargs
-E2E_MODEL = "qwen3.8-27b"
+# enable_thinking false): with thinking, some answers ran to 6-8k tokens,
+# minutes each, and stretched a test from 3.5 to 20 minutes.
+E2E_MODEL = "qwen3.8-27b-nothink"
 
 
 def e2e_roles() -> RoleMap:
@@ -59,16 +61,14 @@ class FreshRepo:
     push_key: Path  # this repository's deploy key, private half
 
 
-@contextlib.contextmanager
+def make_repo(
-def fresh_repo(
+    operator: GiteaClient, name: str, keys: Path, *, agent: str, seed: bool = True
-    operator: GiteaClient, run: str, keys: Path, *, agent: str, seed: bool = True
+) -> FreshRepo:
-) -> Iterator[FreshRepo]:
+    """A repository as the operator would make a project's (D4, D8): seeded
-    """A repository of its own for one test run, as the operator would make a
+    with a small Python package (or only a README), the ``agent`` account a
-    project's (D4, D8): seeded with a small Python package, the ``agent``
+    collaborator with write access, the default branch protected (pushes only
-    account a collaborator with write access, the default branch protected (pushes only by the
+    by the owner, an approval to merge), and a deploy key of its own, written
-    owner, an approval to merge), and a deploy key of its own, written under
+    under ``keys``. A failure part-way deletes what was made."""
-    ``keys``. Deleted afterwards, unless NISSE_E2E_KEEP_REPOS is set."""
-    name = f"nisse-e2e-{run}"
     full = f"{OWNER}/{name}"
     operator._request(
         "POST",
@@ -145,8 +145,35 @@ def fresh_repo(
                 "required_approvals": 1,
             },
         )
-        yield FreshRepo(full, key)
+        # The labels the agent reads (D1), as a project would have them.
-    finally:
+        for label, color in (
-        if not os.environ.get("NISSE_E2E_KEEP_REPOS"):
+            ("bug", "#ee0701"),
+            ("discuss", "#0e8a16"),
+            ("design", "#5319e7"),
+        ):
+            operator._request(
+                "POST", f"/repos/{full}/labels", json={"name": label, "color": color}
+            )
+    except BaseException:
+        delete_repo(operator, full)
+        raise
+    return FreshRepo(full, key)
+
+
+def delete_repo(operator: GiteaClient, full: str) -> None:
     with contextlib.suppress(GiteaError):
         operator._request("DELETE", f"/repos/{full}")
+
+
+@contextlib.contextmanager
+def fresh_repo(
+    operator: GiteaClient, run: str, keys: Path, *, agent: str, seed: bool = True
+) -> Iterator[FreshRepo]:
+    """A repository of its own for one test run (make_repo), deleted
+    afterwards unless NISSE_E2E_KEEP_REPOS is set."""
+    repo = make_repo(operator, f"nisse-e2e-{run}", keys, agent=agent, seed=seed)
+    try:
+        yield repo
+    finally:
+        if not os.environ.get("NISSE_E2E_KEEP_REPOS"):
+            delete_repo(operator, repo.name)

tests/integration/test_end_to_end_loop.py

@@ -239,6 +239,7 @@ class Operator:
         return self.transport.outbound[-1].text if self.transport.outbound else ""
 
 
+@pytest.mark.smoke
 def test_the_slice_end_to_end(
     running: tuple[RunningLoop, InMemoryTransport, str],
 ) -> None:
@@ -501,6 +502,7 @@ def test_a_two_milestone_feature(
         operator.tidy(issues, prs)
 
 
+@pytest.mark.smoke
 def test_a_conversation_becomes_filed_work(
     running: tuple[RunningLoop, InMemoryTransport, str],
 ) -> None:
@@ -636,6 +638,7 @@ def test_a_conversation_becomes_a_new_project(
         assert any(f.startswith("docs/plans/") for f in files), files
 
 
+@pytest.mark.smoke
 def test_a_question_leaves_the_restatement_up(
     running: tuple[RunningLoop, InMemoryTransport, str],
 ) -> None:

uv.lock

@@ -12,6 +12,7 @@ members = [
 [manifest.dependency-groups]
 dev = [
     { name = "pytest", specifier = ">=8" },
+    { name = "pytest-xdist", specifier = ">=3" },
     { name = "ruff", specifier = ">=0.6" },
 ]
 
@@ -46,6 +47,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/d1/d6/3965ed04c63042e047cb6a3e6ed1a63a35087b6a609aa3a15ed8ac56c221/colorama-0.4.6-py2.py3-none-any.whl", hash = "sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6", size = 25335, upload-time = "2022-10-25T02:36:20.889Z" },
 ]
 
+[[package]]
+name = "execnet"
+version = "2.1.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/bf/89/780e11f9588d9e7128a3f87788354c7946a9cbb1401ad38a48c4db9a4f07/execnet-2.1.2.tar.gz", hash = "sha256:63d83bfdd9a23e35b9c6a3261412324f964c2ec8dcd8d3c6916ee9373e0befcd", size = 166622, upload-time = "2025-11-12T09:56:37.75Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ab/84/02fc1827e8cdded4aa65baef11296a9bbe595c474f0d6d758af082d849fd/execnet-2.1.2-py3-none-any.whl", hash = "sha256:67fba928dd5a544b783f6056f449e5e3931a5c378b128bc18501f7ea79e296ec", size = 40708, upload-time = "2025-11-12T09:56:36.333Z" },
+]
+
 [[package]]
 name = "h11"
 version = "0.16.0"
@@ -177,6 +187,19 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/24/25/1de2678b631f5a49215c6c96fff41ba892b0a34df68d6d80292b1b48aa7f/pytest-9.1.1-py3-none-any.whl", hash = "sha256:37a86b45efb9a47a61a36449063e8e18d0cab3161329fc099eb21783169c4f0c", size = 386536, upload-time = "2026-06-19T10:58:31.347Z" },
 ]
 
+[[package]]
+name = "pytest-xdist"
+version = "3.8.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "execnet" },
+    { name = "pytest" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/78/b4/439b179d1ff526791eb921115fca8e44e596a13efeda518b9d845a619450/pytest_xdist-3.8.0.tar.gz", hash = "sha256:7e578125ec9bc6050861aa93f2d59f1d8d085595d6551c2c90b6f4fad8d3a9f1", size = 88069, upload-time = "2025-07-01T13:30:59.346Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ca/31/d4e37e9e550c2b92a9cbc2e4d0b7420a27224968580b5a447f420847c975/pytest_xdist-3.8.0-py3-none-any.whl", hash = "sha256:202ca578cfeb7370784a8c33d6d05bc6e13b4f25b5053c30a152269fd10f0b88", size = 46396, upload-time = "2025-07-01T13:30:56.632Z" },
+]
+
 [[package]]
 name = "pyyaml"
 version = "6.0.3"