The monitoring of tropical forests requires large amount of satellite images, some being free, as the case of Landsat and CBERS series, and other, better spatial resolution, these being paid. We currently have in IBAMA, approximately 49,000 scenes of RapidEye, corresponding to three annual coverages of the entire territory of Brazil, and added to growing acquisition of Landsat 8 images, provides the environmental analyst a condition never taken previously to monitor Brazil's natural resources. On the one hand we have a large number of images to improve the quality of analysis, on the other hand, the use of all images of a given location becomes impossible to add each image manually in desktop GIS. This paper will show the methodology and the implementation of the catalog on the fly, allowing the environmental analyst, have automatically, all cataloged satellite images of a particular area. The work consists of: 1) provide the minimalistic form of satellite images, ie, with minimal computational resources, mainly the demands of settings on the server, and 2) get in the QGIS automatically images of areas interest. Scripts were made in Python and Shell / Linux to generate the satellite images in the format TMS (Tile Map Server) and the definition files to GDAL_WMS (GDAL Web Map Services). The scripts were made using libraries and GDAL programs, and are available on Github (https://github.com/lmotta/scripts-for-gis). For each satellite scene, we have the structure of TMS format directories and files and corresponding XML file with the definition of GDAL_WMS driver. The scenes are processed in parallel through the parallel utility (http://www.gnu.org/software/parallel/), so the processing time, the server is limited processing capability. The entire set of scripts and tools, makes the process minimalistic therefore have the responsability for each division step, not a burden to maintenance, compared to the development of a processing framework. The scenes were cataloged in a vector layer allowing the consultation of these, and for each type of image (Landsat, CBERS, ...) has its cover layer. Plugin Catalog On the Fly was developed for QGIS, allowing access autamoticamente images using as search index the vector layer coverage. The plugin makes the intersection between the extension of the map view (mapcanvas) with the cover layer, and thus the identification of scenes that are in the area of interest. The cover layer has to have two field, a field with image data and the other with the image address, and may be, the URL of the TMS or a local image. The plugin is the official repository of QGIS as experimental (http://plugins.qgis.org/plugins/catalogotf_plugin/) and also on Github (https://github.com/lmotta/catalog-on-the-fly). The Catalog on the Fly can be seen on Youtube (https://www.youtube.com/watch?v=2S3RlWr0uQg). The work is in progress, where the IBAMA satellite images are being processed, and environmental analysts are testing the plugin.