Lexis: Public Records measured; case retrieval not implemented

What this skill carries — grep references/ for any subject the names below miss: !d=${CLAUDE_SKILL_DIR}; command -v skill-toc >/dev/null 2>&1 && exec skill-toc "$d"; s=$HOME/.claude/skills/plugin-utils/bin/skill-toc; [ -x "$s" ] && exec "$s" "$d"; echo "(skill-toc unavailable: references and scripts are NOT listed here — install the plugin-utils plugin, or start a new session so its bin/ reaches PATH)"

There is no Lexis case-retrieval procedure here. Nobody has driven the Lexis+ case-law UI, so there are no measured selectors, endpoints or steps to follow for it.

Use workflows:westlaw. It is measured end to end and produces a publisher-keyed DOCX with star pagination — the same artifact a Lexis route would be built to produce. If the case is available on Westlaw, the answer to "get it from Lexis" is to get it from Westlaw.

Public Records (measured)

Business records only; the account has no DPPA/GLBA permissible use, so person-level sources are closed. Facts, field ids and the worked example: ${CLAUDE_SKILL_DIR}/references/public-records.md — read it before touching the UI. Entity lookups by filing number: bun ${CLAUDE_SKILL_DIR}/scripts/corp-lookup.ts <in.csv> <out.csv> [--full] (CDP :9222, signed-in tab, resumable; --full adds status date, annual list, filing_history JSON and last default/revocation/reinstatement dates; an out.csv from before that has a different header and is refused). Judgments & liens by company/FEIN: bun ${CLAUDE_SKILL_DIR}/scripts/jnl-lookup.ts <in.csv> <out.csv> [--max-records N] [--no-strict] [--fein-only | --debtor-segment] [--dry-run] (--debtor-segment = Terms-and-Connectors debtor("NAME"); it does NOT narrow: 0 hits for one firm, 1,448 loose hits for another, so do not use it to cut noise; --dry-run prints the queries offline; one row per record, 50 per query by default; filter on debtor afterwards, strict search did not narrow Troika; name+FEIN does not narrow either, a blank company with a fein runs a FEIN-only search, --fein-only ignores the company column — see references/public-records.md).

• Delaware is not covered. The jurisdiction dropdown has 50 states without it; do not look for Delaware status on Lexis.
• Clear the form before every search. It keeps the previous terms, and a stale company name ANDs with a new charter number into zero results.
• Pacing is a hard limit: one search at a time, at least 20 s apart, 100 per run by default and never more than 300 (the script refuses a shorter delay or a larger cap). No bulk export; this is for validation samples, not dataset construction.

About to	Why wrong	Do instead
Write an ad-hoc CDP loop for entity lookups	It skips the form-clearing and pacing the script enforces	Run corp-lookup.ts
Sign in, or solve a CAPTCHA, in the user's Lexis tab	Credentials and sessions are the user's	Stop and report
Search Delaware or treat "no results" as "no such entity"	Delaware is not in coverage; check coverage/CorpFil.html	Report not covered

Case retrieval — what we know, UNVERIFIED

Both flagged because they come from documentation and general knowledge, not from a session:

• Lexis offers delivery formats comparable to Westlaw's, including Word and RTF. Which dialog controls exist, whether star pagination is a separate option, and what OPC layout the export uses are all unknown.
• Lexis is rolling out an MCP connector (Protégé). If it delivers verbatim case text rather than synthesized research output, it may make browser driving unnecessary — but whether it does is unverified, and Thomson Reuters' CoCounsel connector notably does not (see the westlaw skill).

If you implement this

Drive the UI and record what you measure. Do not write a plausible-looking procedure from the Westlaw one by analogy — a wrong procedure for a real UI costs more than this stub does, because it reads as authoritative and fails silently partway through.