We report the serendipitous discovery of a faint (M_UV > −12.2), low-metallicity (Z ∼ 0.02 Z⊙) ionizing source, dubbed T2c, with a spectroscopic redshift of z = 6.146. T2c is part of a larger structure amplified by the Hubble Frontier Field galaxy cluster MACSJ0416 and was observed with the James Webb Space Telescope (JWST) NIRSpec integral field unit. Stacking the short-wavelength NIRCam data reveals no stellar continuum detection down to a magnitude limit of m_UV ≃ 31.0 (3σ). However, prominent Hβ, [O III]λλ4959, 5007, and Hα emissions are detected, with equivalent widths exceeding 200 Å, 800 Å, and 1300 Å (3σ), respectively. The corresponding intrinsic (magnification-corrected ×23 ± 3) ultraviolet and optical rest-frame magnitudes exceed 34.4 and 33.9 (corresponding to M_UV and M_opt fainter than −12.2 and −12.8 at λ_rest ∼ 2000 Å and ∼5000 Å, respectively), suggesting a stellar mass lower than a few 10⁴ M_⊙ under an instantaneous burst scenario. The inferred ionizing photon production efficiency (ξ_ion) is high: ξ_ion ≳ 26.08(25.86) 3(5)σ, assuming no dust attenuation and no Lyman continuum leakage. This indicates the presence of massive stars despite the low mass of the object. The very poor sampling of the initial mass function in such a low-mass star-forming complex suggests that the formation of very massive stars might be favored in very low-metallicity environments. T2c is surrounded by Balmer and weak oxygen emission on a spatial scale of a few hundred parsecs, after correcting for lensing effects. This system resembles a H II region potentially powered by currently undetected, extremely efficient, low-metallicity star complexes or clusters. We propose that massive O-type stars populate these low-mass, low-metallicity, high-redshift satellites, likely observed in an early and short formation phase, and contribute to the ionization of the surrounding medium.