fix(agent): fail closed on revision implementer error; refresh wire snapshot

- The wire handshake snapshot (tests_e2e/test_wire_protocol.py) was stale:
  this PR added the judge-minimum-diff and judge-overengineering-review
  bundled skills but never regenerated the handshake skills list, leaving
  CI red on test_initialize_handshake / test_initialize_external_tool_conflict.
  Regenerated to include both new skills.

- ImplementAndJudge: an implementer error on the revision now fails closed to
  BLOCKED, clearing the prior revision's stale NEEDS_WORK verdict and artifact
  so they can't leak into the final result (mirrors the judge-error branch).
  Adds a regression test.

Author: elkaix

src/pythinker_code/tools/agent/__init__.py

@@ -1450,7 +1450,13 @@ async def __call__(self, params: ImplementAndJudgeParams) -> ToolReturnValue:
             )
             last_implementer_output = self._child_result_output(impl_result)
             if impl_result.is_error:
+                # Fail closed: an implementer error on a revision must not let the
+                # prior revision's NEEDS_WORK verdict or artifact leak into the
+                # final result. Reset to BLOCKED, mirroring the judge-error branch.
                 last_implementer_error = impl_result.message
+                last_verdict = "BLOCKED"
+                last_verdict_raw = None
+                last_artifact = None
                 break
 
             last_artifact = _extract_coding_artifact(last_implementer_output)

tests/core/test_implement_judge_chain.py

@@ -374,6 +374,26 @@ async def test_chain_needs_work_hits_cap(runtime: Runtime, monkeypatch: pytest.M
     assert [c[0] for c in calls].count("implementer") == 2
 
 
+async def test_chain_revision_implementer_error_resets_to_blocked(
+    runtime: Runtime, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """An implementer error on the revision fails closed to BLOCKED — the prior
+    revision's NEEDS_WORK verdict and artifact must not leak into the result.
+    """
+    tool, _calls = _make_chain(
+        runtime,
+        monkeypatch,
+        [_ok(_ARTIFACT_OUTPUT), _ok(_JUDGE_NEEDS_WORK), _err("implementer exploded on revision")],
+    )
+    with tool_call_context("ImplementAndJudge"):
+        result = await tool(ImplementAndJudgeParams(brief="do x"))
+    assert result.is_error is True
+    assert result.extras is not None and result.extras["verdict"] == "BLOCKED"
+    assert "implementer exploded on revision" in result.output
+    # The superseded rev-0 artifact must not be presented as the current one.
+    assert "coding_artifact: (missing" in result.output
+
+
 async def test_chain_implementer_error_blocks(
     runtime: Runtime, monkeypatch: pytest.MonkeyPatch
 ) -> None:

tests_e2e/test_wire_protocol.py

@@ -155,6 +155,16 @@ def test_initialize_handshake(tmp_path) -> None:
                             "description": "Implement one or more checked-in specs using scout-plan-implement-verify workflow.",
                             "aliases": [],
                         },
+                        {
+                            "name": "skill:judge-minimum-diff",
+                            "description": "Reduction-ladder and minimum-diff checks the Pythinker judge subagent applies to every non-trivial diff.",
+                            "aliases": [],
+                        },
+                        {
+                            "name": "skill:judge-overengineering-review",
+                            "description": "Over-engineering review checklist the parent runs before declaring non-trivial code changes done.",
+                            "aliases": [],
+                        },
                         {
                             "name": "skill:pr-walkthrough",
                             "description": "Produce a concise reviewer-friendly walkthrough of a PR or diff, including changed areas, behavior, tests, and risks.",
@@ -370,6 +380,16 @@ def test_initialize_external_tool_conflict(tmp_path) -> None:
                             "description": "Implement one or more checked-in specs using scout-plan-implement-verify workflow.",
                             "aliases": [],
                         },
+                        {
+                            "name": "skill:judge-minimum-diff",
+                            "description": "Reduction-ladder and minimum-diff checks the Pythinker judge subagent applies to every non-trivial diff.",
+                            "aliases": [],
+                        },
+                        {
+                            "name": "skill:judge-overengineering-review",
+                            "description": "Over-engineering review checklist the parent runs before declaring non-trivial code changes done.",
+                            "aliases": [],
+                        },
                         {
                             "name": "skill:pr-walkthrough",
                             "description": "Produce a concise reviewer-friendly walkthrough of a PR or diff, including changed areas, behavior, tests, and risks.",