perf(news): scope first-stage Skill prompts to each task

backend/tests/test_news_effort.py

@@ -4,30 +4,32 @@ from types import SimpleNamespace as N
 import pytest
 
 
-@pytest.mark.parametrize('stage', ['index', 'research'])
+@pytest.mark.parametrize('stage', ['index', 'assign', 'research'])
 @pytest.mark.parametrize('selected', [None, 'low', 'medium', 'high', 'xhigh', 'max'])
 @pytest.mark.parametrize('inherited', [None, 'medium'])
 def test_stage_effort_isolated(trading, stage, selected, inherited):
     from alphapulse.models import AiProvider, SkillNode
     from alphapulse.news_research.service import model_config
+    from alphapulse.news_research.contracts import task_skill
+    config_stage = "index" if stage == "assign" else stage
     shared = N(id=1, model='shared', reasoning_effort=inherited, max_tokens=8192, context_window=200000)
     node = N(id=1, scope='news', kind='skill', enabled=True, parent_id=None, published_version_id=1)
-    version = N(version_no=1, content='Synthetic rules')
+    version = N(version_no=1, content='## 索引规则\nSynthetic index\n## 归属规则\nSynthetic assignment')
     class Session:
         async def get(self, model, identifier):
             if model is AiProvider: return N(**vars(shared))
             return node if model is SkillNode else version
         def expunge(self, obj): pass
     config = {key: {'provider_id': 1, 'model': key+'-model', 'skill_id': 1,
-                   'reasoning_effort': selected if key == stage else None} for key in ['index', 'research']}
+                   'reasoning_effort': selected if key == config_stage else None} for key in ['index', 'research']}
     async def check():
         actual, skill, rules = await model_config(Session(), config, stage)
-        assert actual.model == stage+'-model'
+        assert actual.model == config_stage+'-model'
         assert actual.reasoning_effort == (selected if selected is not None else inherited)
         assert rules['model_config']['selected_reasoning_effort'] == selected
         assert rules['model_config']['reasoning_effort'] == actual.reasoning_effort
-        assert rules['content'] == version.content and skill.endswith(version.content)
-        other = 'research' if stage == 'index' else 'index'
+        assert rules['content'] == version.content and skill.endswith(task_skill(version.content, stage))
+        other = 'research' if config_stage == 'index' else 'index'
         assert (await model_config(Session(), config, other))[0].reasoning_effort == inherited
         assert shared.model == 'shared' and shared.reasoning_effort == inherited
     asyncio.run(check())

backend/tests/test_news_research_workspace.py

@@ -681,3 +681,49 @@ def test_long_event_compaction_uses_final_research_context(trading, monkeypatch,
     assert not reader.fits(provider, skill, versions[0].input, ResearchResult)
     assert not reader.fits(provider, boundary_rules, compressed, ResearchResult)
     assert reader.fits(provider, skill, compressed, ResearchResult) != needs_compaction
+
+
+@pytest.mark.parametrize("task", ["index", "assign", "research"])
+def test_model_config_sends_only_selected_task_and_records_exact_prompt(trading, monkeypatch, task):
+    import asyncio
+    from unittest.mock import AsyncMock
+    from alphapulse.news_research import service
+    from alphapulse.news_research.contracts import INDEX_ASSIGN_SKILL, INDEX_SKILL, ASSIGN_SKILL, RESEARCH_SKILL
+    from alphapulse.news_research.pipeline import BASE
+    from alphapulse.models import SkillNode
+
+    content = RESEARCH_SKILL if task == "research" else INDEX_ASSIGN_SKILL
+    node = N(id=22, scope="news", kind="skill", enabled=True, parent_id=None, published_version_id=61)
+    version = N(version_no=5, content=content)
+    session = N(get=AsyncMock(side_effect=lambda model, *args: node if model is SkillNode else version))
+    provider = N(id=12, model="test", max_tokens=1000, context_window=100000, reasoning_effort="medium")
+    monkeypatch.setattr(service, "provider_for", AsyncMock(return_value=provider))
+    config = {stage: dict(provider_id=12, model="test", skill_id=22) for stage in ("index", "research")}
+    _, prompt, rules = asyncio.run(service.model_config(session, config, task))
+    selected = {"index": INDEX_SKILL, "assign": ASSIGN_SKILL, "research": RESEARCH_SKILL}[task]
+    assert prompt == BASE + "\n" + selected
+    assert rules["system_prompt"] == prompt and rules["content"] == content
+    assert rules["skill_id"] == 22 and rules["version"] == 5
+
+
+@pytest.mark.parametrize("content", [
+    "legacy unstructured rules",
+    "## 索引规则\nonly one section",
+    "## 索引规则\n\n## 归属规则\nassign",
+    "## 索引规则\nindex\n## 索引规则\nagain",
+    "preamble\n## 索引规则\nindex\n## 归属规则\nassign",
+    "## 索引规则\nindex\n## 归属规则\nassign\n## 其他规则\nextra",
+    "## 索引规则\nindex\n## 归属规则\nassign\n## 索引规则\nagain",
+])
+def test_invalid_first_stage_skill_fails_instead_of_sending_full_text(trading, content):
+    from alphapulse.news_research.contracts import task_skill, SkillFormatError
+    for task in ("index", "assign"):
+        with pytest.raises(SkillFormatError):
+            task_skill(content, task)
+
+
+def test_task_skill_keeps_custom_rules_and_accepts_crlf(trading):
+    from alphapulse.news_research.contracts import task_skill
+    content = "## 归属规则\r\n自定义归属\r\n### 细则\r\n保留规则\r\n## 索引规则\r\n自定义索引"
+    assert task_skill(content, "assign") == "自定义归属\n### 细则\n保留规则"
+    assert task_skill(content, "index") == "自定义索引"

backend/tests/test_news_scenario.py

@@ -92,7 +92,7 @@ def test_published_skill_retains_base_instructions(trading, monkeypatch):
     monkeypatch.setattr(service, 'provider_for', AsyncMock(return_value=provider))
     session = SimpleNamespace(get=AsyncMock(side_effect=[
         SimpleNamespace(id=9, scope='news', kind='skill', enabled=True, parent_id=None, published_version_id=10),
-        SimpleNamespace(version_no=3, content='PUBLISHED'),
+        SimpleNamespace(version_no=3, content='## 索引规则\nPUBLISHED\n## 归属规则\nASSIGN'),
     ]))
     _, skill, rules = asyncio.run(service.model_config(session, {'index': {'provider_id': 7, 'model': 'test', 'skill_id': 9}}, 'index'))
     assert skill == BASE+'\nPUBLISHED' and rules['base'] == BASE