Accept remote history HAL links

Author: SqlRush

docs/cc-100-roadmap.md

@@ -74,7 +74,7 @@ M6 补充：contract `ID` JSON 读取现在接受 JSON number/null，remote hist
 
 M6 补充：remote history response parser 现在会递归解包 `data.session.events`、`data.projectSession.eventConnection`、`conversation`、`remoteHistory` 等 GraphQL/session wrapper，继续复用 `nodes`/`edges[].node` 和 `pageInfo` pagination 解析。
 
-M6 补充：remote history REST/link 风格分页现在接受 `links.next`/`links.previous`/`links.prev`/`links.older` 的字符串 URL 或 `{href,url,uri,link}` 对象，并从 `before_id`、`beforeId`、`cursor`、`pageCursor`、`previousCursor`、`prevCursor`、`beforeCursor`、`olderCursor`、`startCursor`、`endCursor` 等 query 参数提取下一页 before-id。
+M6 补充：remote history REST/link 风格分页现在接受 `links`/`_links` 的 `next`/`previous`/`prev`/`older` 字符串 URL 或 `{href,url,uri,link}` 对象，并从 `before_id`、`beforeId`、`cursor`、`pageCursor`、`previousCursor`、`prevCursor`、`beforeCursor`、`olderCursor`、`startCursor`、`endCursor` 等 query 参数提取下一页 before-id。
 
 M6 补充：remote history REST/link 风格分页现在也接受 RFC/JSON:API 风格的 `links` 数组，按 `rel`/`relation`/`name`/`type` 中的 `previous`/`prev`/`older`/`next` 选择续抓 URL，并从同一组 before/cursor query 参数提取 before-id。
 

docs/claude-code-go-rewrite-plan.md

@@ -297,7 +297,7 @@ test/parity/                 # golden tests against TS/official behavior
 - 本轮补充：prompt history `LogEntry` 读取接受 `sessionID`/`session`/`sessionUuid`/`sessionUUID`/`session_uuid` 作为 session id 别名，current-session-first 历史排序不会因 session 字段拼写不同而失效。
 - 本轮补充：REPL message restore 现在可从用户消息 content blocks、`imagePasteIds` 和 pasted-content metadata 重建 prompt text、`[Image #N]` 引用和 base64 image pasted contents。
 - 本轮补充：Ctrl-S prompt stash 现在保存/恢复 prompt text、cursor 和 pasted-content metadata，空 prompt 下再次触发会恢复 stash，贴近官方 `chat:stash`。
-- 本轮补充：remote history REST/link 风格分页接受 `links.next`/`links.previous`/`links.prev`/`links.older` 的字符串 URL 或 `{href,url,uri,link}` 对象，并从 `before_id`、`beforeId`、`cursor`、`pageCursor`、`previousCursor`、`prevCursor`、`beforeCursor`、`olderCursor`、`startCursor`、`endCursor` 等 query 参数提取续抓 before-id。
+- 本轮补充：remote history REST/link 风格分页接受 `links`/`_links` 的 `next`/`previous`/`prev`/`older` 字符串 URL 或 `{href,url,uri,link}` 对象，并从 `before_id`、`beforeId`、`cursor`、`pageCursor`、`previousCursor`、`prevCursor`、`beforeCursor`、`olderCursor`、`startCursor`、`endCursor` 等 query 参数提取续抓 before-id。
 - 本轮补充：remote history REST/link 风格分页也接受 RFC/JSON:API 风格 `links` 数组，按 `rel`/`relation`/`name`/`type` 中的 `previous`/`prev`/`older`/`next` 选择 continuation URL。
 - 本轮补充：remote history HTTP `Link` header 分页接受 `rel="previous"`/`prev`/`older`/`next` URL，并从同一组 before/cursor query 参数提取续抓 before-id。
 - 本轮补充：sidechain/subagent state loader 接受 `subagent_start`/`agent_start`/`task_start` 和 `sidechain_end`/`subagent_finish`/`agent_finish`/`task_summary` 等 subtype 别名，并归一化 `active`/`success`/`canceled`/`error` 等状态别名。

docs/first-second-parity-audit.md

@@ -150,7 +150,7 @@ M6 progress now includes:
 - `internal/session`: remote-history pagination now accepts token aliases such as `nextPageToken`, `nextToken`, `pageToken`, and `continuationToken` in response fields and link query parameters.
 - `internal/contracts`: ID JSON decoding now accepts JSON numbers/null so remote-history event, message, session, and parent ID aliases can preserve numeric IDs as strings during transcript materialization.
 - `internal/session`: remote-history response parsing now recursively unwraps GraphQL/session containers such as `data.session.events`, `data.projectSession.eventConnection`, `conversation`, and `remoteHistory` before applying `nodes`/`edges[].node` event-list and `pageInfo` pagination parsing.
-- `internal/session`: remote-history link pagination now accepts `links.next`/`links.previous`/`links.prev`/`links.older` string URLs or `{href,url,uri,link}` objects and extracts before/cursor query parameters for continuation.
+- `internal/session`: remote-history link pagination now accepts `links`/`_links` `next`/`previous`/`prev`/`older` string URLs or `{href,url,uri,link}` objects and extracts before/cursor query parameters for continuation.
 - `internal/session`: remote-history link pagination now also accepts RFC/JSON:API-style `links` arrays with `rel` values such as `previous`, `prev`, `older`, and `next`.
 - `internal/session`: remote-history pagination now also accepts HTTP `Link` header URLs with `previous`/`prev`/`older`/`next` rels as continuation cursor fallbacks.
 - `internal/session`: sidechain state loading now accepts legacy subagent/agent/task start and finish subtype aliases, broader sidechain/subagent ID and type fields, summary aliases, and common running/completed/cancelled/failed status aliases.

internal/session/remote_history.go

@@ -168,15 +168,15 @@ func (r *sessionEventsResponse) mergeJSON(data []byte) error {
 		}
 	}
 	for _, name := range []string{
-		"page", "pagination", "page_info", "pageInfo", "paging", "links", "meta", "metadata",
+		"page", "pagination", "page_info", "pageInfo", "paging", "links", "_links", "meta", "metadata",
 		"session", "project_session", "projectSession", "conversation", "remote_history", "remoteHistory",
 		"event_page", "eventPage", "session_history", "sessionHistory",
 	} {
 		value, ok := raw[name]
 		if !ok {
 			continue
 		}
-		if name == "links" {
+		if name == "links" || name == "_links" {
 			if err := r.mergeLinksField(value); err != nil {
 				return err
 			}

internal/session/remote_history_test.go

@@ -432,6 +432,40 @@ func TestFetchRemoteHistoryAcceptsLinkArrayCursors(t *testing.T) {
 	}
 }
 
+func TestFetchRemoteHistoryAcceptsHALLinkCursors(t *testing.T) {
+	var seen []url.Values
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		seen = append(seen, r.URL.Query())
+		w.Header().Set("Content-Type", "application/json")
+		switch r.URL.Query().Get("before_id") {
+		case "":
+			_, _ = w.Write([]byte(`{"data":[{"type":"status","session_id":"s","status":"latest"}],"_links":{"self":{"href":"/v1/sessions/s/events?cursor=self"},"prev":{"href":"/v1/sessions/s/events?before_id=evt_hal"}}}`))
+		case "evt_hal":
+			_, _ = w.Write([]byte(`{"data":[{"type":"status","session_id":"s","status":"older"}],"_links":{"older":{"href":"/v1/sessions/s/events?pageToken=evt_hal_old"}}}`))
+		case "evt_hal_old":
+			_, _ = w.Write([]byte(`{"data":[{"type":"status","session_id":"s","status":"oldest"}],"_links":{}}`))
+		default:
+			t.Fatalf("unexpected before_id = %q", r.URL.Query().Get("before_id"))
+		}
+	}))
+	defer server.Close()
+
+	authCtx := NewRemoteHistoryAuthContext("s", "token", "", auth.OAuthConfig{BaseAPIURL: server.URL})
+	events, err := FetchRemoteHistory(context.Background(), server.Client(), authCtx, RemoteHistoryFetchOptions{Limit: 3})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !events.Complete || events.Pages != 3 || len(events.Events) != 3 || events.NextBeforeID != "" {
+		t.Fatalf("events = %#v", events)
+	}
+	if events.Events[0].Status != "latest" || events.Events[1].Status != "older" || events.Events[2].Status != "oldest" {
+		t.Fatalf("events = %#v", events.Events)
+	}
+	if len(seen) != 3 || seen[1].Get("before_id") != "evt_hal" || seen[2].Get("before_id") != "evt_hal_old" {
+		t.Fatalf("queries = %#v", seen)
+	}
+}
+
 func TestFetchRemoteHistoryAcceptsLinkHeaderCursors(t *testing.T) {
 	var seen []url.Values
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {