Skip to contents

A lazy dplyr::tbl() over the llm_results table: rows produced by LLM jobs (summaries, analyses, ...) linked to their target via target_id plus target_type (asset, item, collection or unknown). The id is deterministic: llr-{target_type}-{target_id}-{job_type}. The result column is JSON-in-TEXT, parsed by entropia_collect() into a list-column.

Usage

entropia_llm_results(con, target_type = NULL, job_type = NULL)

Arguments

con

A connection returned by entropia_connect().

target_type

Optional character vector of target types to keep, or NULL (default) for all.

job_type

Optional character vector of job types to keep, or NULL (default) for all.

Value

A tbl_sql on llm_results.

Details

Two optional filters, both pushed down to SQL:

  • target_type: keep only rows whose target is one of the given types. On databases that predate migration 0019 the column does not exist and passing a filter errors with guidance; without a filter the accessor still reads the table.

  • job_type: keep only rows for one or more job types.

Examples

con <- entropia_connect(system.file("extdata", "entropia-example.sqlite",
  package = "entropiaR"
))
entropia_collect(entropia_llm_results(con))
#> # A tibble: 1 × 6
#>   id             target_id target_type job_type result       created_at         
#>   <chr>          <chr>     <chr>       <chr>    <list>       <dttm>             
#> 1 llr-item-2222… 22222222… item        summary  <named list> 2026-01-15 12:11:40
entropia_collect(entropia_llm_results(con, target_type = "item"))
#> # A tibble: 1 × 6
#>   id             target_id target_type job_type result       created_at         
#>   <chr>          <chr>     <chr>       <chr>    <list>       <dttm>             
#> 1 llr-item-2222… 22222222… item        summary  <named list> 2026-01-15 12:11:40
entropia_disconnect(con)