An award-winning transparency tool that combines over 50 official NYC open datasets — agencies, people, jobs, capital projects, schools, districts and procurement — into one searchable picture of how city government works.
This pull request replaces the text-based "child_of" column with a formal "parent_org_id" foreign key constraint in the organizations table. It resolves three live bugs regarding parent organization display in the MCP tools and chatbot, deletes an obsolete PHP script, and adds 14 new tests to prevent regression.
Added a detailed migration plan document outlining the steps to move WeGov GitHub repositories to the Sarapis organization. The plan triages 24 repositories, identifies production remote configurations, and highlights security cleanup tasks. The new planning document was also added to the public sync exclusion list to protect sensitive details.
This update restricts the token-set similarity guard to trigger only when a sync would write a different value, preventing false-positive human review flags. It also updates the project documentation to mark Phases 2 and 4b as complete and bumps the normalizer-py submodule pointer.
This update automates the normalizer's organization matching dictionary by deriving it directly from the live register instead of using a frozen snapshot. It introduces a new API endpoint for name variants, adds a curated aliases table to handle edge cases, implements a collision policy for duplicate names, and updates the registry refresh script with pre-flight checks.
Updated the orgs matching recipe to fetch data directly from the databook API instead of a non-existent local JSON file. This change automates organization directory updates, preserves name variant mapping, and adds JSON response unwrapping.
This pull request schedules the organization registry refresh script to run weekly with automated health monitoring and Sentry error reporting. It secures the unattended process by wrapping the crosswalk rebuild in a transaction with a minimum row guard, fixing a bug that caused incorrect link rejections, and adding post-run validation checks.