Works with the AI Agents you already use
claim severity segmentation
Raw claim data often hides the true drivers of loss costs.
$1.99
claim severity segmentation
Example session with this skill installed
Analyze these 2023 property claims from the attached CSV. Group them by claim_type and show the severity distribution using standard bands. Compute percentiles for the total incurred amounts.
- Read your context and instructions
- Compiled the claim severity segmentation
Analysis of the 2023 property claims dataset is complete. I have processed the records through the severity segmentation script to compute distributions and cohort statistics.
Claim Severity Segmentation — Property Portfolio, 2023-10-27
1. Overall Statistics
- Total Claims: 142
- Total Incurred: $1,845,200
- Mean Incurred: $12,994
- Median Incurred: $4,200
- Standard Deviation: $42,150
- 25th Percentile: $1,150
- 75th Percentile: $12,800
- 95th Percentile: $85,000
2. Severity Distribution
| Severity Band | Count | Percentage | Mean Incurred | Median Incurred |
|---|---|---|---|---|
| Minimal (<$1k) | 45 | 31.7% | $420 | $380 |
| Low ($1k-$5k) | 58 | 40.8% | $2,850 | $2,900 |
| Moderate ($5k-$20k) | 22 | 15.5% | $11,200 | $10,500 |
| High ($20k-$100k) | 12 | 8.5% | $52,000 | $48,000 |
| Severe ($100k-$500k) | 4 | 2.8% | [Suppressed < 5] | [Suppressed] |
| Catastrophic (>$500k) | 1 | 0.7% | [Suppressed < 5] | [Suppressed] |
3. Analysis by Claim Type
- Fire/Lightning: 12 claims, Mean $84,500. Shows highest severity skew.
- Water Damage: 85 claims, Mean $4,200. High frequency, low severity.
- Theft: 45 claims, Mean $2,100. Consistent low-band concentration.
4. Cohort Analysis
- Q1 2023: 34 claims, Mean $11,500.
- Q2 2023: 41 claims, Mean $14,200.
- Q3 2023: 67 claims, Mean $13,100.
5. Observations
- Water damage represents the bulk of frequency but remains in the Low severity band.
Analytical estimate — validation required.
- The 95th percentile is heavily driven by three outlier Fire claims exceeding $100k.
Analytical estimate — validation required.
- Median cost is significantly lower than the Mean, indicating a right-skewed distribution common in property lines.
Analytical estimate — validation required.
Next steps
- Review the individual claim files for the "Severe" and "Catastrophic" bands to verify reserve accuracy.
- Compare these 2023 results against 2022 historical data to identify severity inflation.
- Validate the Fire/Lightning cause codes to ensure consistent categorization.
Connects securely to your tools. The creator never sees your data.
About this skill
Raw claim data often hides the true drivers of loss costs. Manual calculation of severity bands and percentiles is error-prone and time-consuming for insurance analysts. This skill automates the segmentation of historical claims to reveal cost distributions and cohort patterns.
What it does
- Statistical Computation calculates mean, median, standard deviation, and percentiles for incurred claim amounts.
- Severity Segmentation groups claims into bands from minimal to catastrophic based on cost.
- Cohort Analysis compares severity metrics across different years, quarters, or custom groupings.
- Type Breakdown isolates severity patterns by specific claim types and cause codes.
- Data Validation automatically excludes claims without paid amounts and suppresses small samples for privacy.
How it works
- Data Extraction identifies claim IDs, amounts, dates, and types from your provided documents or datasets.
- Request Generation builds a structured JSON input following a strict schema to ensure calculation accuracy.
- Deterministic Processing executes a specialized Python script to compute the distribution without manual intervention.
- Report Composition generates a standardized report with tables covering overall stats, bands, and observations.
Frameworks & tools
This skill utilizes a deterministic Python execution environment for statistical processing. It relies on JSON schema validation for input integrity and follows specific actuarial-aligned grouping methods.
Why this beats prompting it yourself
Generic prompts often struggle with accurate percentile math and consistent banding. This skill uses a hardcoded script to ensure your mean, median, and standard deviation are mathematically sound every time, preventing the "hallucinated math" common in standard LLM outputs.
Use cases
- Analyze historical claim costs to identify high-severity loss drivers.
- Segment a claims portfolio into standardized cost bands for reporting.
- Compare claim severity between different underwriting years or business cohorts.
- Audit claim type distributions to see which causes lead to catastrophic costs.
Known limitations
This tool is for historical analysis only. It does not perform loss development, IBNR estimation, or reserve adequacy forecasting. It requires at least 5 claims per group to report statistics.
Not for predictive modeling, IBNR estimation, or reserve adequacy. Requires at least 5 claims per group for privacy. Does not handle claims missing paid_amount.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago