Help us improve
Share bugs, ideas, or general feedback.
From retriever
Guides first-pass legal review using Retriever: intake documents, find hot docs, triage contracts, build privilege sets, and export review sets.
npx claudepluginhub sdemyanov/retrieverHow this skill is triggered — by the user, by Claude, or both
Slash command
/retriever:legal-reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> Operates under `retriever:routing`. If the user's intent actually fits a different tier — another `retriever:*` skill, a Tier 2 slash, a Tier 3 `tools.py` subcommand, or (last resort) direct DB access — stop and re-route against the ladder before continuing.
Hands off high-volume clause extraction to Luminance or Kira and performs quality assurance on the output based on configured trust levels. Useful for bulk review of standardized contracts.
Reviews contracts, MSAs, SOWs, and NDAs against internal context from Slack, Notion, and email searches. Delivers startup-focused risk assessments.
Routes inbound agreements (vendor, NDA, SaaS) to the correct review skill based on document structure and integrates results into a single memo.
Share bugs, ideas, or general feedback.
Operates under
retriever:routing. If the user's intent actually fits a different tier — anotherretriever:*skill, a Tier 2 slash, a Tier 3tools.pysubcommand, or (last resort) direct DB access — stop and re-route against the ladder before continuing.
Use this skill as a thin wrapper over Retriever's core review engine. The user
should not have to learn /search, /filter, /scope, or export syntax
before they can do useful legal work.
When the user is new or the next step is unclear, start with the simplest Retriever instruction:
Good examples:
retriever:workspace + retriever:ingestretriever:ingest-productionretriever:pstretriever:workspace + retriever:ingest for
persistent review; use live Drive access only for quick one-off questionsretriever:search, retriever:dataset, retriever:filter,
retriever:bates, retriever:scope, and paging/sort controls.retriever:field and
retriever:fill, but suggest new custom fields before creating them unless
the request clearly implies field creation.privilege_status, hot_doc,
issue_tag, and review_notes.retriever:run-job when a processing run is more appropriate than ad hoc
browsing.retriever:export.