GitHub lists 7 projects: https://github.com/search?utf8=%E2%9C%93&q=toxics+release+inventory&type= But the EPA site seems only to offer JSON for UV index, which i guess means how likely you'll get a sunburn? https://www.epa.gov/enviro/web-services#uvrest Data.gov shows a REST API which is a bit more helpful, i think: https://catalog.data.gov/dataset/toxic-release-inventory-tri-8162e/resource/8585d7d4-a2bb-4d22-994e-1a61a852ee1d

Oh, and maybe Skytruth Alerts is a good place to go? But I'm not sure it does historical data: http://alerts.skytruth.org/ They do have an API: http://alerts.skytruth.org/api.php But it's not clear to me if these overlap with TRI reports?

Oh, cool -- this page (https://www.epa.gov/enviro/envirofacts-data-service-api) shows this type of query, starting with a zip code prefix: https://iaspub.epa.gov/enviro/efservice/tri_facility/zip_code/BEGINNING/0290/JSON ...for facilities in zip codes starting with 0290_ for example. That shows 86 facilities.

www.scorecard.org was an early, but now defunct step towards that. also just my usual reminder that while EJ groups have been able to mobilize TRI data for righteous ends it is industry self-reported data, so it should be taken with a grain of salt. Here's a paragraph i wrote about related database: An environmental lawyer interviewed in the course of this fieldwork found that the Pennsylvania Department for Environmental Protection’s (DEP) estimation of airborne emissions resulting from unconventional natural gas compressor sites to be “impossibly low.” Not just under-estimated but “several orders of magnitude lower than they could possibly achieve or that I had ever seen anywhere in any permit or any stack test anywhere.” As suggested by this investigator, the air quality described by official records of environmental pollution was a world beyond chemical and mechanical feasibility. This specific assertion of annual industry emissions emerged out of a compendium of 300 spreadsheets with multiple tabs. An environmental advocacy organization employed a programmer to write a python script to convert this unwieldy dataset into a comprehensible format. Without such an investigation, the multiple realities enacted by this data might have remained one more unquestioned monitoring point. https://citizensense.net/indeterminate-data-monitoring-extraction-sites/

Here's another option. https://toxmap.nlm.nih.gov/toxmap/