Multi-agent legal memo drafting plugin for Claude Cowork and Claude Code — intake, research, review, polished .docx output
Agent
Searches relevant case law (judgments, decisions, opinions) for issues from a memo plan. Routes US case law to CourtListener MCP and multi-jurisdictional case law to Legal Data Hunter MCP. Structures findings as prevailing / conflicting / recent positions.
Agent
Audits citations in a legal memo draft against the research files that produced it. Verifies every normative/case/doctrine claim is grounded in research, paraphrase matches the source, and currency-blocking issues are respected. Reads draft AND research files.
Agent
Independent clarity review of a legal memo draft. Checks sentence length, jargon-without-explanation, accessibility for a non-lawyer business stakeholder. Reads only the draft.
Agent
Final external-client readiness review before docx export. Checks tone, assumptions, disclaimers, confidentiality, recommendation quality, and whether the memo is shippable with minimal manual edits.
Agent
Stress-tests a legal memo draft by finding contrary authority, overconfident conclusions, missing caveats, and ways an opposing lawyer or regulator would attack the analysis.
Agent
Verifies that the sources collected by researchers are still current law — checks cross-references between acts, status of cited judgments, age of doctrinal guidance, and reachability of URLs. Produces a blocking/non-blocking report for the writer.
Agent
Searches doctrinal sources, soft law, regulatory guidance (EDPB, national DPAs, etc.), and peer-reviewed academic commentary for memo issues. Activated only when plan.md indicates Doctrine yes.
Agent
Performs preliminary legal triage before the main research plan. Identifies missing facts, legal variables the user may not know to provide, safe default assumptions, and must-answer intake questions.
Agent
Independent logical-coherence review of a legal memo draft. Checks IRAC structure, premise-conclusion soundness, inter-issue consistency. Reads only the draft, isolated from research and prior reviews. Returns structured JSON.
Agent
Writes (v1) or rewrites (vN) the legal memorandum based on research files, the selected template, and (for vN) mediator's consolidated revision instructions. Produces structured markdown with IRAC analysis per issue and full source citations.
Agent
Reviews completed legal research before drafting. Checks whether each planned issue has sufficient primary sources, contrary authority, currency-sensitive sources, and explicit gaps.
Agent
Consolidates parallel reviewer JSONs (3 in Brief mode, 5 in Full) into a single actionable revision list for the writer. Reads the configured reviewer set from state.json.config.reviewerlist. Resolves reviewer conflicts via house-style priority order. Updates state.json with exit decision.
Agent
Builds an evidence table from research and currency files so the writer and citation auditor work from a structured source pack rather than loose notes.
Agent
Searches primary normative acts (statutes, regulations, directives, secondary legislation) for issues listed in a memo plan. Routes queries to Legal Data Hunter MCP first; uses official WebFetch fallback for US eCFR/govinfo and other primary portals when needed. Returns structured findings grouped by issue.
Agent
Extracts a custom style+formatting profile from example memos and/or written rules. Produces prose-style.md (always) and template.md (when structural input is present), plus meta.json. Called only by the style skill when the user creates a new profile.
Agent
Independent style review of a legal memo draft. Detects AI-tells, em-dash overuse, inflated symbolism, AI vocabulary, vague attributions, grammar/punctuation issues. Reads only the draft.