prompts

prompts/

MCP prompt implementations — structured workflow templates that guide LLM clients through multi-step legal tasks.

Prompts vs. tools: tools are executable actions that return data; prompts are guidance templates that tell an LLM how to approach a task and which tools to chain. They are returned to the client as message sequences, not executed server-side.

Modules and prompts

Module	Prompts
research_prompts.py	precedent_analysis — step-by-step guide for researching and applying case precedents; statutory_interpretation — framework for reading a statute's text, structure, and legislative history
drafting_prompts.py	brief_construction — scaffolded workflow for drafting a legal brief from issue to conclusion; clause_comparison — structured approach to comparing contract clause alternatives
analysis_prompts.py	contract_review — risk-triage workflow for full contract analysis using analyze_clauses and generate_negotiation_guide; citation_validation — formatting cleanup plus authoritative source-checking workflow
argument_prompts.py	argument_development — IRAC-based argument scaffold for motions and briefs; authority_integration — methodology for weaving case law and statutes into a coherent argument
__init__.py	register_all_prompts(mcp) — wires all four prompt modules into the FastMCP instance

The 8 prompts at a glance

Prompt name	One-line description
precedent_analysis	Research and apply relevant case precedents to a legal issue
statutory_interpretation	Interpret a statute using text, structure, and legislative history
brief_construction	Scaffold a complete legal brief from issue statement to conclusion
clause_comparison	Compare contract clause alternatives and select the lower-risk option
contract_review	Full contract risk triage with negotiation guidance
citation_validation	Format and normalize citations, then route existence and good-law checks to authoritative sources
argument_development	Build an IRAC-structured argument for a motion or brief
authority_integration	Integrate case law and statutes into a coherent legal argument

Registration

register_all_prompts(mcp) in __init__.py is called unconditionally from main.py — prompts are not gated by feature flags.

Adding a new prompt

1. Create prompts/my_prompts.py with a register_my_prompts(mcp) function.
2. Decorate each prompt with @mcp.prompt() and return a list of Message objects (or a string that FastMCP wraps automatically).
3. Import and call register_my_prompts(mcp) from prompts/__init__.py.

# prompts/my_prompts.py
def register_my_prompts(mcp):

    @mcp.prompt()
    def my_workflow(context: str) -> str:
        """Guide the LLM through a custom legal workflow."""
        return (
            f"You are analyzing: {context}\n"
            "Step 1: Call search_precedents to find relevant cases.\n"
            "Step 2: Call extract_statute for applicable statutes.\n"
            "Step 3: Synthesize findings and surface the disclaimer."
        )