SKILL.md
Overview
The Weight Tracker database uses a JSON file to store weight entries provided by the user.
File Location
The data file is available at {root}/db/weight-tracker.json
Writing to the db
- Before writing into the db, identify which is the best approach:
- in-memory change: load the entire file in-memory, apply the change, serialize it back - text-based change: scan the file in text mode and apply the change as a patch
- Isolate the atomical change you want to apply
- Read existing data to understand which kind of operation is needed
- insert - update - delete
- Apply the change with the identified approach.
ALWAYS: explain the execution plan to the user as output in the chat
Reading from the DB
- Identify the best approach to the read request:
- in-memory change: load the entire file in-memory, evaluate logic to perform the read
- text-based change: scan the file in text mode and calculate the expected information as context
- Perform the read of the pertinent information
- Perform aggregations if needed
ALWAYS: explain the execution plan to the user as output in the chat
Data Structure
Always ensure the database keeps a data structure like this:
{
"2025": {
"11": {
"10": 99.0,
"12": 99.5
}
}
}More info:
- the indexing of the data is: year -> month -> day
- the value of a day is a number, the database is unit agnostic
- if a new writing request targets an existing day, override the value for the day