Contract Risk Intelligence

A Python contract-review tool with an interactive dashboard.

It reads contract PDFs, sends the extracted text to Claude, scores each contract for business risk, flags risky clauses, and saves the output to JSON and CSV.

Then the dashboard turns those results into a clear executive view.

Know the risk before you sign.

---

What This Project Does

• Reads every PDF inside the contracts/ folder.
• Extracts contract text with pdfplumber.
• Sends each contract to the Claude API for structured risk analysis.
• Produces detailed results in results.json.
• Produces a spreadsheet summary in results.csv.
• Shows the results in a single-page dashboard: index.html.
• Keeps API keys private through .env and .gitignore.

---

Dashboard

The dashboard is built as one self-contained HTML file:

index.html

It loads:

• results.csv for the contract register and KPI counts.
• results.json for detailed summaries, red flags, and clause-level findings.

Dashboard views include:

• Executive summary
• Risk distribution
• Priority contracts
• Recurring risk areas
• Full contract register
• Contract detail drawer

Risk tiers are standardized and transparent:

Low    0-39
Medium 40-69
High   70-100

The dashboard also derives the displayed tier from the score, so the visual counts stay consistent with the data.

---

Risk-Tier Logic

The risk score comes from Claude's contract analysis.

The final risk tier is controlled by Python.

This means the model does not decide the final tier label. The script applies fixed thresholds after the score is returned:

Low    = 0-39
Medium = 40-69
High   = 70-100

This keeps results.json, results.csv, and the dashboard aligned.

Current reconciled sample counts:

Tier	Count
Low	7
Medium	11
High	7
Total	25

---

Files

File	Purpose
analyze_contracts.py	Main Python script that extracts text, calls Claude, applies risk-tier thresholds, and writes outputs
index.html	Interactive dashboard for reviewing the results
results.json	Full detailed analysis for each contract
results.csv	Flattened spreadsheet summary
requirements.txt	Python dependencies
.env.example	Shows the required API key format
.gitignore	Keeps .env, virtual environments, and cache files out of Git
contracts/	Folder for PDF contracts

---

Requirements

You need:

• Python 3
• An Anthropic API key
• Contract PDFs

Install the Python packages:

pip3 install -r requirements.txt

---

Add Your API Key

Create a file named .env in the project folder:

ANTHROPIC_API_KEY=your_key_here

Replace your_key_here with your real Anthropic key.

Do not commit .env.

It is already ignored by Git.

---

Run The Contract Analyzer

Place PDF contracts in:

contracts/

Then run:

python3 analyze_contracts.py

The script will create or update:

results.json
results.csv

---

Run The Dashboard Locally

Because the dashboard loads local CSV and JSON files, open it through a local server.

From the project folder:

python3 -m http.server 8011

Then open:

http://localhost:8011/index.html

Do not double-click index.html.

Most browsers block local file loading from file://.

---

Security Notes

• The real API key belongs only in .env.
• .env is ignored by Git.
• .env.example shows the format without exposing secrets.
• Sample contracts are fictional and used for testing.

---

Current Status

This project includes:

• A working Python analyzer
• Standardized risk-tier logic
• Reconciled JSON and CSV outputs
• A polished single-page dashboard
• 25 sample contracts for testing

Built by Acey Magallanes as a practical AI automation portfolio project.