From 4734e206a9187752df68720976d5263cc597b4bd Mon Sep 17 00:00:00 2001
From: Debug Agent <debug@example.com>
Date: Wed, 20 May 2026 11:10:15 +0200
Subject: [PATCH] perf(adapter): opt into ?include_skills=false on batch
 conversation fetch
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Canvas only reads ``agent.kind`` and ``agent.llm.model`` from
``GET /api/conversations`` responses (see ``toAppConversation``) — the
~260 KB ``agent.agent_context.skills`` payload that
``load_user_skills=true`` / ``load_public_skills=true`` agents
accumulate is never accessed client-side, but every conversation
poll has to parse + structurally-share it, which is the bottleneck
the perf profiling identified.

Pass ``{ includeSkills: false }`` to ``ConversationClient.getConversations``
so the agent-server trims the payload at the route boundary. The
local agent-server runtime keeps the full skill catalog server-side
for prompt rendering — pure client-side perf with no functional
change.

Companion PRs (both pinned via SHA temporarily so this PR is
end-to-end runnable / lintable / profileable; SHAs flip to released
versions once both companions ship):

- software-agent-sdk#3316: server-side ``include_skills`` query param
- typescript-client#172: typed ``includeSkills`` option on
  ``ConversationClient``, pinned via the SHA in package.json

Measured impact (full empirical profile in
software-agent-sdk#3301): ~260 KB → ~5 KB on the wire,
``first event painted`` 2226 ms → 1283 ms (-42%) for a 40-skill
conversation cold-open.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 package-lock.json                                   |  4 ++--
 package.json                                        |  2 +-
 .../agent-server-conversation-service.api.ts        | 13 ++++++++++++-
 3 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/package-lock.json b/package-lock.json
index 744f57f12..350121d69 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -12,7 +12,7 @@
         "@heroui/react": "2.8.10",
         "@microlink/react-json-view": "1.31.20",
         "@monaco-editor/react": "4.7.0",
-        "@openhands/typescript-client": "git+https://github.com/OpenHands/typescript-client.git#v0.6.0",
+        "@openhands/typescript-client": "git+https://github.com/OpenHands/typescript-client.git#6aa5593b0a0fba0be25881706c84feebd9802a40",
         "@react-router/node": "7.14.2",
         "@react-router/serve": "7.14.2",
         "@tailwindcss/vite": "4.2.4",
@@ -3440,7 +3440,7 @@
     },
     "node_modules/@openhands/typescript-client": {
       "version": "0.1.1",
-      "resolved": "git+https://github.com/OpenHands/typescript-client.git#ef62e82fc3dfb03991a1c8025429caf354427263",
+      "resolved": "git+https://github.com/OpenHands/typescript-client.git#6aa5593b0a0fba0be25881706c84feebd9802a40",
       "license": "MIT",
       "dependencies": {
         "@openrouter/sdk": "^0.12.35",
diff --git a/package.json b/package.json
index 89774e43a..98a3418c9 100644
--- a/package.json
+++ b/package.json
@@ -23,7 +23,7 @@
     "@heroui/react": "2.8.10",
     "@microlink/react-json-view": "1.31.20",
     "@monaco-editor/react": "4.7.0",
-    "@openhands/typescript-client": "git+https://github.com/OpenHands/typescript-client.git#v0.6.0",
+    "@openhands/typescript-client": "git+https://github.com/OpenHands/typescript-client.git#6aa5593b0a0fba0be25881706c84feebd9802a40",
     "@react-router/node": "7.14.2",
     "@react-router/serve": "7.14.2",
     "@tailwindcss/vite": "4.2.4",
diff --git a/src/api/conversation-service/agent-server-conversation-service.api.ts b/src/api/conversation-service/agent-server-conversation-service.api.ts
index 0a20b18cb..e541465f8 100644
--- a/src/api/conversation-service/agent-server-conversation-service.api.ts
+++ b/src/api/conversation-service/agent-server-conversation-service.api.ts
@@ -411,9 +411,20 @@ class AgentServerConversationService {
       return batchGetCloudConversations(ids);
     }
 
+    // Opt into ``include_skills=false`` so the agent-server trims the
+    // ~260 KB ``agent.agent_context.skills`` payload from each
+    // conversation. Canvas only reads ``agent.kind`` and
+    // ``agent.llm.model`` from these responses (see ``toAppConversation``
+    // — the skills list is never accessed), and the local agent-server
+    // runtime keeps the full skill catalog server-side for prompt
+    // rendering. Companion PRs:
+    //   - software-agent-sdk#3316 (server-side query param)
+    //   - typescript-client#172 (typed ``includeSkills`` option)
     const data = await new ConversationClient(
       getAgentServerClientOptions(),
-    ).getConversations<DirectConversationInfo>(ids);
+    ).getConversations<DirectConversationInfo>(ids, {
+      includeSkills: false,
+    });
 
     return requireDirectConversationItems(data).map((item) =>
       toAppConversation(item),