Sustainability

The problem

Embodied carbon is increasingly regulated and incentivized — LEED v4.1 MRc2, Buy Clean policies, and net-zero commitments all require Environmental Product Declarations. But EPDs are dense PDFs with inconsistent formats, scattered across dozens of registries, and the data needs to flow into specifications and whole-building LCA.

How it works

Four skills form a pipeline: parse EPD PDFs into structured data, search registries for published EPDs, compare products on environmental impact metrics, and generate CSI specification sections with GWP thresholds.

Skills

Skill	What it covers
/epd-parser	Extract structured data from EPD PDFs — GWP, life cycle stages, certifications
/epd-research	Search EC3, UL, Environdec, and manufacturer sites for EPDs by material
/epd-comparator	Side-by-side impact comparison with LEED MRc2 eligibility check
/epd-to-spec	CSI specification sections requiring EPDs and setting GWP limits

Pipeline

EPD PDFs ──→ /epd-parser ──→ structured data ←── /epd-research ←── registries
                                    │
                             /epd-comparator
                                    │
                          comparison report
                                    │
                              /epd-to-spec
                                    │
                         CSI spec sections

Each skill works standalone. Chaining is natural but not required.

Output

All skills share a 42-column EPD schema covering product identity, EPD metadata, impact indicators (GWP-fossil, GWP-biogenic, ODP, AP, EP, POCP), resource use, and LEED eligibility. Data saves to CSV or Google Sheets. Comparison reports and specifications save as markdown.