Research Ingest
Copy the block below into a new Claude Code session and add a file path on the last line. Self-contained orientation for ingesting case law, statutes, regulations, or search results into any active research project.
When to use: after running a vLex / Fastcase / Westlaw query or downloading a bulk export, drop the resulting file path into a fresh Claude session with this block. Claude will detect the file type, infer the project and jurisdiction, and run the appropriate workflow.
When NOT to use: ad-hoc legal questions (just ask), NCSC AI Toolkit Atlas writes (use atlas-harvester skill), or anything outside the research-ingest pattern.
Paste block
[Research ingest] I'm continuing legal research work. Before doing anything: read MEMORY.md to identify which research project this file belongs to. Active research projects are indexed there (Charter Data Privacy is current; others may exist). Infer the project from filename and content; ask once if unclear. For the inferred project, follow the standard ingest workflow: 1. Detect file type (CSV search results, ZIP bulk download, single .doc, .pdf). 2. Infer jurisdiction from filename or content (federal, ny, nyc, ct, ma, etc.). 3. If CSV: read it, triage by relevance tier (Tier 1 bullseyes, Tier 2 strong, Tier 3 noise), identify which hits are bullseyes for the project, queue any new case citations into the project's Case-Law-Verification-Queue. Recommend a capture batch with rank positions. 4. If ZIP: extract into the right jurisdiction subfolder (create if new). Separate genuine new wins from duplicates against existing captures. 5. If single .doc or .pdf: file into the right jurisdiction folder with a normalized filename. For case PDFs, file under Primary-Source-PDFs/Case-Law/. For statute PDFs, under Primary-Source-PDFs/Statutes/. 6. Update the jurisdiction's README.md with any new statutes, regs, agency guidance, or case citations. 7. Update the project's lib/data.js if structural data changed (new law, new jurisdiction, verified case moving from caseQueue to verifiedCases). 8. If lib/data.js changed: cd to the project's app folder, git add -A, commit with descriptive message (author jcolarusso@gmail.com), git push to master, then npx --yes vercel deploy --prod --yes. 9. ALWAYS sync Legal Brain after markdown updates: cd C:\Users\joe\dev\legal-brain-app, npm run sync, git add -A, git commit -m "Sync: <what was added>", git push origin master. Vercel auto-deploys legal-brain via its build command (which runs sync first). The Legal Brain app at legal-brain-app.vercel.app is the master markdown viewer for the entire MKL Legal brain and must stay current alongside any project-specific apps. 10. Report: files added, deduped count, verification queue items resolved, what's still pending. Always confirm BOTH apps are updated (project app + Legal Brain). Hard rules: - Never invent case citations; queue as pending until vLex-verified with a URL or local PDF. - Commit author MUST be jcolarusso@gmail.com (Vercel deploy gate). - Direct master push on jsmg-am repos is the default. - vLex read-only posture; no CTS or client doc uploads to vLex. - No em dashes in chat replies or in Joe-voiced text. - Match existing markdown and data.js schemas; do not invent new ones. File to ingest: [PATH]
Active research projects
- Charter Data Privacy (this app) — federal + NY + NYC layeredMKL/Legal/02_Data-Privacy/Charter-Data-Privacy-Research/
More projects will appear here as they're created. The block adapts automatically: Claude reads MEMORY.md to find them.
Apps that stay in sync
- charter-data-privacy-appStructured viewer (this app)
- legal-brain-appMarkdown viewer for all of MKL/Legal/ (auth: joe/legalbrain). Synced via `npm run sync` after any markdown change.
What Claude will do
- Detect file type and route to the right research project
- Infer jurisdiction from filename or content
- Triage CSV / extract ZIP / file single docs
- Update the project's markdown research
- Update the project's app data.js if structural change
- Commit + deploy the project app if data.js changed
- Always sync Legal Brain after markdown updates, then commit + push (Vercel auto-deploys via its build script)
- Report status for BOTH apps
Tips
- For a NEW research project, mention that explicitly so Claude scaffolds the folder + data.js entry.
- For case-law VERIFICATION (you ran a vLex query and got the case PDF), drop the PDF or ZIP and Claude will move the entry from caseQueue to verifiedCases.
- Bulk downloads from Fastcase / vLex come as ZIPs with a contents .html and the case PDFs. Claude extracts only the PDFs by default.