Generate structured, deduplicated tracker data from research inputs with provenance.
npx skills add https://github.com/GYF0311/lorekit --skill data-research-gyf0311
Please help me install this Agent Skill. Skill: data-research Source: https://github.com/GYF0311/lorekit/tree/main/brain/skills/data-research Command: npx skills add https://github.com/GYF0311/lorekit --skill data-research-gyf0311
Data arrives across emails, web pages, and attachments, but tracking it accurately requires repetitive searching, extraction, deduplication, and keeping canonical tracker pages up to date without double-counting.
Ask your AI to run the data-research skill to build or update a canonical tracker from investor update emails using the investor-updates recipe.
High-intent search queries and answers about installing and using this skill.
To convert investor update emails into structured tracker data, you can use a YAML-defined recipe to search sources, classify relevance, extract fields, and archive raw provenance. This process applies recurring pipelines to maintain canonical markdown trackers.
Deduplicating entity tracking data requires skipping exact matches, flagging fuzzy matches for review, and recording corrections when amounts differ for the same entity and date. This ensures accurate canonical tracker updates without double-counting.
Maintaining a canonical tracker table involves updating the correct markdown sections, computing running totals, and backlinking entities to ensure cross-referenced knowledge. YAML-defined recipes automate this recurring email-to-tracker workflow.
Data extraction requires a deterministic-first processing approach with an LLM fallback to maintain extraction integrity. The system ensures accuracy by re-reading saved files, linking raw sources, and applying canonical tracker updates with deduplication rules.
You can use YAML-defined recipes for any email-to-tracker workflow, including donations and company metrics. The structured research pipeline searches sources, classifies items, extracts fields, and archives raw inputs for provenance across various recurring pipelines.
When amounts differ for the same entity and date, the deduplication process flags these fuzzy matches for review and records corrections. This prevents double-counting while maintaining accurate canonical tracker data with backlinked cross-references.
ltdrewGather sources and extract fields into canonical tracker pages.
semicolon-devteamAggregate open bug reports from repositories into consolidated JSON or Markdown reports.
ngochuy13Convert research inputs into structured, deduplicated tracker entries with raw-source archiving.
PostergullyAutomate structured data research from emails, web sources, and APIs.
