AI-powered tools for FDA medical device regulatory work. Built for regulatory affairs professionals working on 510(k) submissions.
From your terminal:
claude plugin marketplace add andrewlasiter/fda-predicate-assistant
claude plugin install fda-predicate-assistant@fda-toolsOr from inside a Claude Code or Claude Desktop session:
/plugin marketplace add andrewlasiter/fda-predicate-assistant
/plugin install fda-predicate-assistant@fda-tools
Start a new session after installing to load the plugin.
- Find predicates — Search FDA databases, trace predicate lineage, manually propose predicates, and validate device numbers
- Analyze safety — Pull MAUDE adverse events, recall history, and UDI/GUDID records for any product code or device
- Look up standards — Search FDA Recognized Consensus Standards by product code, standard number, or keyword
- Plan your submission — Get pathway recommendations, generate testing plans, and prepare Pre-Sub packages
- Generate documents — Substantial equivalence tables, submission outlines, regulatory prose drafts for 18 eSTAR sections
- Run calculators — Shelf life (ASTM F1980 accelerated aging), sample size, and sterilization dose calculations
- Assemble for filing — eSTAR-structured packages, import/export eSTAR XML, traceability matrices, consistency checks
- Simulate FDA review — Pre-check your submission against RTA checklists and identify likely deficiencies before filing
- Maintain your data — Gap analysis, automated PDF downloads, predicate extraction, and merge pipeline
- Run it all at once — Full autonomous pipeline from extraction through SE comparison
- Monitor changes — Watch for new clearances, recalls, MAUDE events, and guidance updates
35 commands, 4 autonomous agents, and 712 tests covering every stage of the 510(k) workflow — from predicate research through CDRH Portal submission. Integrates with all 7 openFDA Device API endpoints and bundles Python scripts for batch PDF processing and data pipeline maintenance.
See the full documentation for commands, agents, installation details, and quick start examples.
MIT