Skip to content
Back to skills

Legal 5step Framework

ASecurity

BetterCallClaude 5-step Swiss legal framework — an end-to-end pipeline that chains five agents sequentially: (1) doc-analyze for intake and fact extraction, (2) swiss-legal-research for precedent and statute lookup via BGE/ATF/DTF, fedlex-sparql, entscheidsuche, and onlinekommentar MCP servers, (3) swiss-legal-strategy for claim strength, risk matrix, and settlement evaluation, (4) adversarial-analysis for three-agent Erwägung-style stress test, (5) swiss-legal-drafting for verified document ...

  • 39 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added May 28, 2026
ai-agents

Works with

  • cli
  • mcp

Security analysis

A100/100

Scanned May 28, 2026

npx -y skills add fedec65/bettercallclaude --skill legal-5step-framework --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Legal 5step Framework?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Legal 5step Framework
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/fedec65-legal-5step-framework/badge)](https://www.skillsdirectory.com/skills/fedec65-legal-5step-framework)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: legal-5step-framework
description: "BetterCallClaude 5-step Swiss legal framework — an end-to-end pipeline that chains five agents sequentially: (1) doc-analyze for intake and fact extraction, (2) swiss-legal-research for precedent and statute lookup via BGE/ATF/DTF, fedlex-sparql, entscheidsuche, and onlinekommentar MCP servers, (3) swiss-legal-strategy for claim strength, risk matrix, and settlement evaluation, (4) adversarial-analysis for three-agent Erwägung-style stress test, (5) swiss-legal-drafting for verified document production. Trigger when: a user invokes /bettercallclaude:legal-5step, asks to run the full framework on a case, wants a complete legal analysis from intake to draft, or says 'run all steps', 'full pipeline', 'end-to-end analysis', or 'complete legal workflow'. Do NOT trigger for: single-step tasks (use the individual command: research, strategy, adversarial, draft, or doc-analyze), citation-only work (use swiss-citation-formats), quick legal questions (use swiss-legal-research directly), or translation only (use swiss-legal-translation). Boundary with /bettercallclaude:workflow: the workflow command executes predefined pipelines (litigation-prep, due-diligence, etc.); legal-5step is a fixed five-step framework that always runs the same sequence and is optimized for repeatable client intake."
---

# BetterCallClaude 5-Step Legal Framework

You coordinate the BetterCallClaude 5-step Swiss legal pipeline. Five agents run sequentially, each passing structured output to the next, from raw case input through a verified drafted legal document.

## Framework

```
INTAKE → RESEARCH → STRATEGY → ADVERSARIAL → DRAFT
  (1)       (2)        (3)          (4)         (5)
```

The five steps map to the five cognitive moves on any new Swiss legal matter:
1. What are the facts and issues?
2. What does the law say?
3. What should we do?
4. Are we right?
5. How do we say it?

## Step Definitions

### Step 1: INTAKE
**Agent**: doc-analyzer | **Skill**: `swiss-jurisdictions`, `privacy-routing`

Extracts facts, identifies legal issues, determines jurisdiction and language, flags Anwaltsgeheimnis markers before any external MCP call. Privilege flag propagates through Steps 2–5.

**Output**:
- Structured fact list
- Legal issues ranked by relevance
- Jurisdiction: federal or cantonal (XX)
- Language: DE/FR/IT/EN
- Privilege flag: true/false

### Step 2: RESEARCH
**Agent**: researcher | **Skill**: `swiss-legal-research`  
**MCP servers**: `swiss-caselaw`, `bge-search`, `entscheidsuche`, `fedlex-sparql`, `onlinekommentar`

Retrieves controlling BGE/ATF/DTF precedents, live statute text, and scholarly commentary. All citations generated via `swiss-caselaw:cite` — never constructed manually.

**Output**:
- Research memorandum with verified citations
- Statute text retrieved live from `fedlex-sparql` or `swiss-caselaw:get_law`
- Doctrine from `onlinekommentar`

**Citation rule**: Every BGE/ATF/DTF reference in Steps 3–5 must appear in this memorandum.

### Step 3: STRATEGY
**Agent**: strategist + risk-analyst | **Skill**: `swiss-legal-strategy`  
**MCP tools**: `entscheidsuche:analyze_precedent_success_rate`, `entscheidsuche:find_similar_cases`

Converts the research memo into a decision-oriented recommendation with precedent-grounded probability.

**Output**:
- Claim strength: Strong / Moderate / Weak
- Success probability: X% (grounded in `analyze_precedent_success_rate`)
- Procedural pathway: ZPO track, forum, timeline
- Risk matrix: Critical / Moderate / Minor
- Settlement range and BATNA
- Recommendation: litigate / settle / ADR

**Checkpoint**: If `success_probability < 30%` or any Critical risk — pause, present memo, await confirmation.

### Step 4: ADVERSARIAL
**Agent**: advocate + adversary + judicial-analyst | **Skill**: `adversarial-analysis`  
**MCP tools**: `swiss-caselaw:find_citations`, `swiss-caselaw:get_erwaegung`, `bge-search`

Stress-tests the Step 3 strategy through three independent agents before drafting commits the position to paper.

**Output**:
- AdvocateReport: strongest case FOR, with BGE support
- AdversaryReport: systematic challenge with counter-precedents
- JudicialSynthesis: Erwägung-style balanced conclusion with probability scores
- strategy_delta: % change from Step 3 estimate

**Checkpoint**: If `strategy_delta > 15%` — pause before Step 5, present both estimates, invite strategy revision.

### Step 5: DRAFT
**Agent**: drafter + citation-specialist | **Skill**: `swiss-legal-drafting`, `swiss-citation-formats`  
**MCP tools**: `swiss-caselaw:cite`, `legal-citations` (format validation)

Produces the legal document from the confirmed position. No new legal arguments introduced — everything traces to Step 2.

**Document selection**:

| Matter Type | Default Output |
|-------------|---------------|
| Litigation — plaintiff | Klageschrift (ZPO Art. 221) |
| Litigation — defendant | Klageantwort (ZPO Art. 222) |
| Contract dispute | Rechtsgutachten |
| Compliance | Advisory memorandum |
| Advisory | Client letter or memo |
| Custom | As specified |

**Citation formats by language**:

| Language | BGE | Article | Paragraph |
|----------|-----|---------|-----------|
| DE | BGE 147 III 265 E. 3.1 | Art. 97 Abs. 1 OR | Abs. |
| FR | ATF 147 III 265 consid. 3.1 | art. 97 al. 1 CO | al. |
| IT | DTF 147 III 265 consid. 3.1 | art. 97 cpv. 1 CO | cpv. |
| EN | BGE 147 III 265 at 3.1 | Art. 97(1) OR | para. |

## Data Flow

```
Step 1  →  facts, issues, jurisdiction, language, privilege_flag
Step 2  →  research memo, verified citations, statute text, doctrine
Step 3  →  strategy memo, success_probability, risk_matrix, recommendation
Step 4  →  judicial synthesis, overall_probability, strategy_delta
Step 5  →  final document (all citations from Step 2 memo)
```

## Quality Gates

| Gate | Condition | Action |
|------|-----------|--------|
| Privilege | `privilege_flag: true` | Pause before Step 2, confirm MCP calls |
| Low probability | `success_probability < 30%` | Pause after Step 3 |
| Strategy delta | `strategy_delta > 15%` | Pause before Step 5 |
| Citation integrity | Citation in Step 5 not in Step 2 memo | Block and fetch via MCP |

## Professional Disclaimer

All pipeline outputs require review and validation by a qualified Swiss lawyer before use. This framework does not constitute legal advice. Lawyers maintain full professional responsibility for all legal work products.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.