different-ai/openwork · error · Error

serializeSDKError(result.error)

Error message

serializeSDKError(result.error)

What it means

When a slash-command draft is submitted, the UI calls opencodeClient.session.command() against the primary endpoint. The SDK returns a result envelope; a non-null result.error means the server rejected the command. The UI serializes that SDK error (serializeSDKError, session-route.tsx:254) into a thrown Error so it surfaces in the normal send-error path.

Source

Thrown at apps/app/src/react-app/shell/session-route.tsx:1341

                    label: modelSelection,
                  },
                ];
                trackSessionActive(targetSessionId, telemetryDimensions);
                trackTaskStarted(targetSessionId, telemetryDimensions);

                if (draft.mode === "shell") {
                  await shellInSession(opencodeClient, targetSessionId, text);
                  return;
                }

                if (draft.command) {
                  const result = await opencodeClient.session.command({
                    sessionID: targetSessionId,
                    command: draft.command.name,
                    arguments: draft.command.arguments,
                  });
                  if (result.error) {
                    throw new Error(serializeSDKError(result.error));
                  }
                  return;
                }

                const parts = await draftToParts(draft, selectedWorkspaceRoot, targetSessionId, selectedWorkspaceEndpoint);
                const envSystemContext = await buildOpenworkEnvSystemContext(client, {
                  cacheKey: targetSessionId,
                  runtimeKey: environmentRuntimeKey,
                });
                const result = await opencodeClient.session.promptAsync({
                  sessionID: targetSessionId,
                  parts,
                  model: sendModel ?? undefined,
                  agent: selectedAgent ?? undefined,
                  ...(sendVariant ? { variant: sendVariant } : {}),
                  ...(envSystemContext ? { system: envSystemContext } : {}),
                });
                if (result.error) {

View on GitHub (pinned to 2b7df46e8a)

Solutions

1. Read the serialized error text for the server's reason (unknown command vs. execution failure)
2. Verify the command name exists in the workspace's command configuration
3. Check the session still exists and retry; start a new session if it was deleted
4. Update the app if the client and server versions are mismatched

const known = await opencodeClient.command.list();
if (!known.data?.some((c) => c.name === draft.command.name)) {
  notify(`Unknown command: ${draft.command.name}`);
  return;
}

try {
  const result = await opencodeClient.session.command({ sessionID, command, arguments });
  if (result.error) throw new Error(serializeSDKError(result.error));
} catch (e) {
  showErrorToast(String((e as Error).message));
}

When it happens

Trigger: Submitting a draft with draft.command set where opencodeClient.session.command() returns { error } — unknown command name, command failed server-side, invalid arguments, or session no longer exists.

Common situations: Typing a custom command that the server's command list no longer defines; passing wrong-typed arguments to a command; session was reverted/deleted between draft and submit; server version mismatch after an app update.

Related errors

• app.error_command_not_resolved
• jsonrpc_error

AI-assisted analysis of different-ai/openwork@2b7df46e8a (2026-09-01). Data as JSON: /api/errors/100537b62d021e24. Report an issue: GitHub.