```

feat(learning): 添加技能学习候选者合成锁定机制

添加了 DraftSynthesisInProgress 和 DraftHasNoChanges 异常来处理并发场景，
确保同一技能学习候选者的合成过程不会重复执行。实现了 claim_learning_candidate_for_synthesis
方法来原子性地锁定候选者进行合成。

fix(web): 为技能草案创建端点添加适当的HTTP状态码

当草案没有变化或正在合成时，现在正确返回409状态码而不是内部错误。

feat(skills): 实现技能修订内容比较以检测无变化情况

添加了 _is_noop_revision 方法来比较基础技能和提议的修订，
如果内容没有实际变化则抛出 NoDraftChanges 异常。

refactor(process): 修复任务证据记录后根运行状态更新逻辑

将任务证据记录事件后的状态从 waiting 更改为 done，并设置 finished_at 时间戳。

feat(tools): 防止在同一运行中重复执行外部写入操作

为邮件发送、日历创建等外部写入工具添加去重机制，避免重复的外部操作。

test: 添加技能学习和工具执行的单元测试

增加测试用例验证并发草案合成、重复外部写入抑制和无变化修订检测等功能。
```

app-instance/backend/beaver/skills/learning/service.py

@@ -20,8 +20,9 @@ from beaver.plugins.tree_merge import merge_supporting_file_trees
 from beaver.skills.drafts.service import DraftService
 from beaver.skills.learning.evidence import EvidencePacket, EvidenceSelector
 from beaver.skills.learning.synthesizer import SkillDraftSynthesizer
-from beaver.skills.catalog.utils import parse_frontmatter
+from beaver.skills.catalog.utils import parse_frontmatter, strip_frontmatter
 from beaver.skills.specs import SkillActivationReceipt
+from beaver.skills.specs.serialization import normalize_frontmatter
 
 
 @dataclass(slots=True)
@@ -30,6 +31,10 @@ class RunReceiptContext:
     effect_records: list[SkillEffectRecord] = field(default_factory=list)
 
 
+class NoDraftChanges(ValueError):
+    """Raised when synthesis produces the same effective skill content as the base version."""
+
+
 class SkillLearningService:
     def __init__(
         self,
@@ -231,13 +236,18 @@ class SkillLearningService:
             )
         target_skill = candidate.related_skill_names[0]
         base_version = candidate.evidence.get("skill_version")
+        base_skill = self._base_skill_snapshot(target_skill, base_version)
         payload = await self.synthesizer.synthesize_revision(
             candidate,
             packet,
             provider,
             model,
-            base_skill=self._base_skill_snapshot(target_skill, base_version),
+            base_skill=base_skill,
         )
+        if self._is_noop_revision(payload, base_skill):
+            raise NoDraftChanges(
+                f"Synthesis produced no changes for {target_skill}/{base_version or 'current'}"
+            )
         return self.draft_service.create_revision_draft(
             skill_name=target_skill,
             base_version=base_version,
@@ -340,6 +350,16 @@ class SkillLearningService:
             "tool_hints": list(loaded.version.tool_hints),
         }
 
+    @staticmethod
+    def _is_noop_revision(payload: dict[str, Any], base_skill: dict[str, Any] | None) -> bool:
+        if base_skill is None:
+            return False
+        base_frontmatter = normalize_frontmatter(dict(base_skill.get("frontmatter") or {}))
+        proposed_frontmatter = normalize_frontmatter(dict(payload.get("frontmatter") or {}))
+        base_body = _normalize_skill_body(str(base_skill.get("content") or ""))
+        proposed_body = _normalize_skill_body(str(payload.get("content") or ""))
+        return base_frontmatter == proposed_frontmatter and base_body == proposed_body
+
     def _merged_base_skill_snapshot(self, skill_names: list[str]) -> dict[str, Any] | None:
         snapshots = [
             snapshot
@@ -602,6 +622,10 @@ class SkillLearningService:
         return parsed.astimezone(timezone.utc)
 
 
+def _normalize_skill_body(content: str) -> str:
+    return "\n".join(line.rstrip() for line in strip_frontmatter(content).strip().splitlines()).strip()
+
+
 def _digest_map(root: Path) -> dict[str, dict[str, Any]]:
     digest = hash_plugin_skill_tree(root)
     return {