Length of data from UniProt_get_entry_by_accession exceeds context window of many LLMs

[delalamo]

When running UniProt_get_entry_by_accession, the resulting data is huge and causes some LLMs like Claude Sonnet to crash. For example, if I query the accession for hen egg white lysozyme (UniProt P00698), then the resulting data, when saved as a JSON, exceeds 41,000 lines in length. Would it be possible to restructure those outputs so that the relevant information is retrieved? Alternatively, can warnings be added to the tool description?

[gasvn]

Hi, for now, I have updated the tool description to give a warning. Please let me know if you think there are better strategies. Thank you!

• WARNING: This tool returns the complete UniProtKB entry, which can be extremely large (40,000+ lines for some entries) and may cause LLM context limits to be exceeded or crashes. For most use cases, consider using more specific extraction tools instead: UniProt_get_function_by_accession, UniProt_get_sequence_by_accession, UniProt_get_recommended_name_by_accession, UniProt_get_organism_by_accession, UniProt_get_subcellular_location_by_accession, UniProt_get_disease_variants_by_accession, or UniProt_get_ptm_processing_by_accession.`