Box Legal Workflows Intake

Client Intake & Onboarding

PREREQUISITES:

• Use the box skill for Box MCP auth, tool selection, and base workflows. If it is not installed, run: npx skills add https://github.com/box/skills --skill box
• Use the box-legal-workflows skill for Box collaboration role definitions, Box AI usage boundaries, and reusable confirmation phrasings. If it is not installed, run: npx skills add box/skills --skill box-legal-workflows

Do client intake in Box: inventory the intake folder, check completeness with Box AI against the firm’s checklist, extract intake data into Box metadata, route via comments and collaborations, and generate the engagement letter with Box DocGen. This skill is the intake-specific recipe; the underlying Box tool mechanics live in the capability references below. The firm supplies the required-document checklist and risk criteria; conflict, PEP, and sanctions determinations run through the firm’s screening system — not the agent or Box AI. Not legal advice.

Box capability references

Reach for these for tool mechanics rather than restating them here:

• The box skill’s references/content-workflows.md — inventory the folder, metadata templates, set_file_metadata, file comments
• The box skill’s references/ai-and-retrieval.md — completeness checks and data extraction; pacing, limits, citations
• The box skill’s references/mcp-doc-gen.md — generate the engagement letter from a template
• The box skill’s references/collaboration.md — tag/route the attorney and share with the client
• The box skill’s references/mcp-search.md — find/inspect the firm’s metadata template

Box metadata model

Record the intake outcome as file metadata so submissions stay searchable. Find or create the firm’s template using the box skill’s references/mcp-search.md and its references/content-workflows.md.

• Representative fields (confirm the firm’s actual set): client_name, matter_name, practice_area, matter_owner, jurisdiction, matter_value, intake_status (complete/incomplete), risk_rating, assigned_attorney, decision, decision_date.
• Store the firm/attorney’s rating and decision, not the agent’s.

Tool selection

Workflow

1. Inventory: list_folder_content_by_folder_id. [CONFIRM: intake folder ID]
2. Completeness: ai_qa_multi_file — is each required doc (per the firm’s checklist) present, complete, and valid, with citations? [CONFIRM: firm’s required-document list]
3. Extract intake data: ai_extract_structured_from_fields_enhanced (client_name, matter_name, practice_area, jurisdiction, matter_value, …).
4. Surface risk indicators (facts only): ai_qa_multi_file to surface the facts the firm’s criteria reference (value, jurisdiction, named parties) with citations. Do not use AI to determine conflicts, PEP status, or sanctions — pass names/entities/locations to the firm’s screening system.
5. Persist: set_file_metadata with the firm-confirmed risk_rating, intake_status, and decision.
6. Route: create_file_comment to tag the attorney with the factual summary; create_collaboration for access. [CONFIRM: who, access level]
7. Engagement letter (only if the firm approves): create_docgen_batch against the firm’s template → output to the confirmed folder → share via create_collaboration or add_folder_shared_link. [CONFIRM: DocGen template, destination folder, share method/expiration]

Legal guardrails

Box mechanics (DocGen template/tag requirements, external-sharing confirmation, AI pacing/limits/citations, metadata writes) are governed by the capability references above and box-legal-workflows. Specific to intake:

• Conflict, PEP, and sanctions clearance are firm screening + attorney/compliance calls — the agent surfaces the underlying text only and never clears them.
• Risk rating and routing/staffing are firm policy + attorney calls, never the agent’s.
• Prompt Box AI to surface facts and cite the source, not to determine risk.