Heavy metal ions pose persistent threats to environmental safety and public health, creating an urgent demand for analytical platforms capable of sensitive, reliable, and visual detection. Herein, a polyoxometalate-functionalized covalent organic framework composite (Ru-POCOF) is developed as a trimodal sensing platform integrating fluorescence (FL), UV–vis absorption, and electrochemiluminescence (ECL) for the detection of Cu²⁺, Fe³⁺, and Hg²⁺. In this architecture, the Ru-COOH (Ru[dcbpy]₃²⁺) complex serves as the photoelectrochemical signal center, while aminated Dawson-type P₂W₁₈ clusters are incorporated into an ordered COF scaffold, providing abundant coordination sites and efficient charge-transfer pathways. Benefiting from this synergistic structure, Ru-POCOF exhibits distinct ion-dependent responses across multiple channels: Cu²⁺ and Fe³⁺ induce significant quenching of FL and UV signals accompanied by visible luminescence changes under UV irradiation, whereas Hg²⁺ is sensitively detected through ECL attenuation. Furthermore, machine learning–assisted analysis of spectral features enables accurate discrimination between Fe³⁺ and Cu²⁺, significantly enhancing analytical reliability in complex matrices. In addition to sensing capability, Ru-POCOF demonstrates rapid adsorption toward Fe³⁺ with a removal efficiency exceeding 60% within 20 min. Reliable detection is validated in tap water, rat serum, and Codonopsis pilosula extracts. Overall, this multifunctional POM–COF hybrid integrates multimodal sensing, visual detection, intelligent analysis, and efficient adsorption, providing a versatile strategy for monitoring trace heavy metal ions in complex systems.