Very low magnetic fields are detected up to now using SQUID-based devices. In the femtotesla range, only low-T/sub c/ SQUIDs have a sufficiently low level of noise to measure extremely low signals, as for instance induced fields of the neural activity of the brain, for magneto-encephalography (MEG). An alternative to SQUIDs is to combine a superconducting flux-to-field transformer with a high-sensitivity giant magnetoresistive sensor (GMR). We have fabricated and performed experiments on such an integrated YBCO-based mixed sensor. It is comprised of a GMR stack separated from a YBCO thin film by an insulating layer. The induced supercurrent in the YBCO is forced through a narrow constriction, thereby creating a local field much larger than the external field. This large local field is detected by the GMR. Effective noise levels are down to 30 fT/sqrtHz, which is in the range of high-Tc Squids. Performance of this device is shown from liquid helium temperature (4.2 K) up to liquid nitrogen temperature (77K). By replacing the GMR by a Tunneling Magneto Resistance (TMR) or a Colossal Magneto Resistance (CMR) sensor, and by adjusting the dimensions to optimize the local enhancement effect, the sensitivity should reach the subfemtotesla range at low frequency. A large number of applications could be investigated by this sensor, especially in the biomagnetism area, as well as in various other fields, from earth mapping to fundamental physics (vortex motion in superconductors, magnetic interactions...).