Story
As a user, I should be able to use my preferred ai client to interact with govbot data.
Description
Make govbot accessible directly inside AI chat interfaces like ChatGPT (as a GPT/action) and Claude (as an MCP server or project tool) so users can query legislative data conversationally.
Technical Requirements
Your choice :). Volunteer work is fueled on passion.
Some Interfaces
- experience to select data catalogs to download.
- query data catalogs.
- (future: work with indexers/classifiers)
From there users can use ai tools to do whatever they want it (data analysis, ask questions, etc).
Key considerations
- Which govbot commands were designed for getting newsfeeds from legislation updates, not necessarily querying. Do we need to update this?
- How to handle the data volume (47+ jurisdictions) within token/context limits
- Offline-friendly: could an MCP server run locally against cloned repos?
About The Data
You can learn more about our data structure in our doc site or our official proposal to Open Civic Data
govbot_data
This currently has our 47 legislation jurisdictions, but executive branch and judicial branch are part of the dream.
Think npm but for for trustable data.
Wild dream would be to add all kinds of data catalogs like catalog.data.gov and data.chicago.gov.
bill.json
The original govbot filetree was proposed with a file tree matching the OpenCivicData spec, but theres another folder:
logs
Log are newsfeeds when its of legislation data. By being a folder/file with timestamp prefix names for the json, it makes it very fast to create a buffer stream from newest to oldest.
Think of them as an timestamped index for the datasets.
I do wonder about removing the log folder from the official data catalogs and have govbot build it since its just an index, but i havent spiked on that yet.
Story
As a user, I should be able to use my preferred ai client to interact with govbot data.
Description
Make govbot accessible directly inside AI chat interfaces like ChatGPT (as a GPT/action) and Claude (as an MCP server or project tool) so users can query legislative data conversationally.
Technical Requirements
Your choice :). Volunteer work is fueled on passion.
Some Interfaces
From there users can use ai tools to do whatever they want it (data analysis, ask questions, etc).
Key considerations
About The Data
You can learn more about our data structure in our doc site or our official proposal to Open Civic Data
govbot_data
This currently has our 47 legislation jurisdictions, but executive branch and judicial branch are part of the dream.
Think npm but for for trustable data.
Wild dream would be to add all kinds of data catalogs like catalog.data.gov and data.chicago.gov.
bill.json
The original govbot filetree was proposed with a file tree matching the OpenCivicData spec, but theres another folder:
logs
Log are newsfeeds when its of legislation data. By being a folder/file with timestamp prefix names for the json, it makes it very fast to create a buffer stream from newest to oldest.
Think of them as an timestamped index for the datasets.
I do wonder about removing the log folder from the official data catalogs and have govbot build it since its just an index, but i havent spiked on that yet.