KilnMCP
A FastMCP server and three-agent Python pipeline serving synthetic biochar pyrolysis kiln telemetry with EWMA/k-sigma anomaly detection, root-cause diagnosis prompts and draft dMRV batch reports.
- Period
- October 2026
- Release
- Public source project
- Work type
- Independent open-source project
- Development
- AI-assisted, with verification
What I built
A FastMCP server and three-agent Python pipeline serving synthetic biochar pyrolysis kiln telemetry with EWMA/k-sigma anomaly detection, root-cause diagnosis prompts and draft dMRV batch reports.
Includes the Lampung Dual-Feedstock slow pyrolysis profile (BATCH-LPG-2027-001, 55% coffee shell + 45% corn cob, 450–650 °C target zone, 369.60 kg CO2e per 200 kg dry biochar via stoichiometric 44/12 × 0.80 × 0.90 accounting, and QR bag reconciliation across 2.0 t/ha Lampung Ultisol plots).
Developed with AI coding assistance. Source code, reproduction instructions, the interactive workbench and tests are available via the links above.
What was checked
36/36 pytest passing tests verified locally on 7 November 2026 (including Lampung dual-feedstock batch stoichiometry, QR bag mass reconciliation, UART $KILN,...*XOR frame integrity, and EWMA/k-sigma fault isolation).
Verification date: 7 November 2026. Tests cover the included synthetic fixtures and do not establish physical reactor performance.
What this demonstrates
All kiln telemetry is synthetic and every generated report is labeled dMRV-style DRAFT - NOT CERTIFIED. Does not represent physical reactor measurements or certified carbon credits.
All projects