Skip to content

Feat/issue 15 llm descriptions#36

Open
theshantanujoshi wants to merge 3 commits into
adhit-r:mainfrom
theshantanujoshi:feat/issue-15-llm-descriptions
Open

Feat/issue 15 llm descriptions#36
theshantanujoshi wants to merge 3 commits into
adhit-r:mainfrom
theshantanujoshi:feat/issue-15-llm-descriptions

Conversation

@theshantanujoshi

Copy link
Copy Markdown

This PR addresses Issue #15 by adding a local build script (scripts/build-llm-descriptions.ts) that uses Transformers.js and the Qwen1.5-0.5B-Chat model to enrich entity descriptions in the knowledge base.

Key Changes:

  • New Script: Added scripts/build-llm-descriptions.ts to identify and fill description gaps in the archive using local AI.
  • Workflow Integration: Updated package.json with a new build:llm-descriptions command.
  • Enriched Data: Successfully generated and merged new lore descriptions for several entities (including Beru Lars, Wicket, and Ayla Secura) into data/build/kb.json.
  • Manifest Updates: Updated data/build/manifest.json to track LLM enrichment counts.

Verification:

  • Tested the script locally with npx tsx scripts/build-llm-descriptions.ts --limit 1 --skip-review.
  • Verified that the "Lore" section correctly appears in the Datapad UI for the enriched entities.

@safedep

safedep Bot commented May 13, 2026

Copy link
Copy Markdown

SafeDep Report Summary

Green Malicious Packages Badge Green Vulnerable Packages Badge Green Risky License Badge

No dependency changes detected. Nothing to scan.

Installation is not linked with SafeDep Tenant. Click here to optionally link your GitHub App installation with SafeDep Tenant.

This report is generated by SafeDep Github App

@adhit-r

adhit-r commented May 14, 2026

Copy link
Copy Markdown
Owner

Holding this for cleanup before merge. npm run typecheck passes, but git diff --check fails on trailing whitespace in scripts/build-llm-descriptions.ts.

The bigger review concern is the generated data/build/kb.json churn: this PR changes ~11k lines of generated data alongside the script. Please split the script/tooling change from generated content, or keep the generated output to a small reviewed sample so the data changes are auditable.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants