Patent landscape analysis

Access and scope

Set CRAWLORA_API_KEY to your Crawlora key. Prefer the listed Crawlora MCP tools when connected; otherwise use the bundled scripts/crawlora.sh, which sends x-api-key to https://api.crawlora.net/api/v1. Read reference/endpoints.md for exact methods and parameters. Check application code as well as HTTP status; payloads are inside data. Stop on authentication errors, back off on 429, and retry a transient upstream failure once. Bound requests by the requested scope and credit budget. Preserve source IDs, URLs, source dates, and retrieval/crawl times separately.

Build an exploratory technology landscape from a documented query strategy. Establish technical scope, jurisdictions, date basis/window, document types, assignee scope, and collection budget before presenting counts or trends.

Build a traceable search corpus

1. Use synonyms and classification lookups to refine the technical query. Keep included/excluded terms and each exact query. Google Patents structured dates default to priority date when dates are supplied; explicitly choose date_field=priority, filing, or publication to match the analysis.
2. Google search pages are zero-based. Record total/results/pages, many_results, requested pages, and stopping rules. Its top-assignee/inventor/classification breakdowns describe the full query match set, not just the collected page; do not mix their denominator with your smaller sample or family counts.
3. USPTO uses BRS field/operator syntax rather than Google's structured filters. Match jurisdictions and date meaning when cross-checking. Its returned guid and database feed detail's guid and source; do not synthesize them. The REST helper rejects literal @ in query arguments to prevent local-file shorthand. For BRS @ date expressions use the connected MCP tool, or use Google's explicit date filters; keep the helper's file guard intact.
4. Fetch detail for selected publication numbers to obtain family members, classifications, abstracts/claims, original/current assignees, and citations. Translate only when needed and retain the original/source language context. Search snippets alone cannot support a detailed claim-coverage map.

scripts/crawlora.sh /googlepatents/search q="solid state battery" date_field=publication after=2020-01-01 page=0 num=20
scripts/crawlora.sh /usptoppubs/search q='battery.ti. AND lithium.ab.' num=20 page=0
scripts/crawlora.sh /googlepatents/coverage
# Resolve returned publication numbers and classification codes before detail.

Group families, entities, and technical themes

Keep publication, application, and family counts separate. Group documents only with returned family evidence; shared titles, inventors, or keywords do not establish a patent family. Preserve the chosen grouping rule and all member publications. Do not label a sampled/incomplete family graph exhaustive.

Normalize assignee spelling with an alias ledger while retaining original and current assignees and unresolved entity matches. Subsidiaries and similarly named firms are not automatically one owner. Keep reported ownership/status dated and attributed; the source's GRANT/APPLICATION search status is not a current rights-validity determination. Citations and similar documents are relevance leads, not proof of infringement, technical dependence, quality, or commercial value.

Code technical themes from the available abstract/claims/detail, allowing multiple labels and unknown evidence. Compare counts over equivalent query/jurisdiction/date scopes and distinguish publications from family-level inventions. Coverage totals measure the index's grants/applications by authority/year, not this technology's output or R&D investment. Return the query/coverage ledger, family/assignee map, topic matrix, dated trend observations, sample documents, and missing evidence. This exploratory landscape does not establish freedom to operate, patentability, validity, enforceability, or a complete prior-art search.