Mcp Chainladder
Hands Claude the math to compute actuarial loss reserves through the chain-ladder method. Exposes eight tools over stdio: parse CSV triangles from disk, run volume-weighted or custom-selected development factors, project to ultimate, calculate IBNR by accident year, run Mack (1993) stochastic standard errors with cross-row covariance, and check Mack (1994) model diagnostics including inflation slope, calendar-year effects, and standardized residuals. Results match the textbook 10×10 triangle to twelve decimal places. You can exclude individual cells (outlier treatments), apply tail factors, toggle between incremental and cumulative, and ask Claude to interpret p-values on assumption tests. Pure calculation, not actuarial judgment. Python 3.10+, pipx installable, works in Claude Desktop, Cursor, and Cline. Licensed under standard PyPI terms.
- tools
- 8
- transport
- STDIO
- last updated
- May 14, 2026
Tools
8 toolsPublic tool metadata for what this MCP can expose to an agent.
compute_chain_ladderEnd-to-end chain ladder.
End-to-end chain ladder.
No parameter schema in public metadata yet.
project_triangleFills the lower-right of the triangle with chain-ladder projections.
Fills the lower-right of the triangle with chain-ladder projections.
No parameter schema in public metadata yet.
mack_stochasticMack (1993) distribution-free standard errors.
Mack (1993) distribution-free standard errors.
No parameter schema in public metadata yet.
mack_diagnosticsReturns standardised residuals, outliers, the calendar-year sign test, Spearman independence test, and the inflation slope of mean log-link-ratio against accident-year index.
Returns standardised residuals, outliers, the calendar-year sign test, Spearman independence test, and the inflation slope of mean log-link-ratio against accident-year index.
No parameter schema in public metadata yet.
parse_csv_triangleReads a CSV from disk, treats blank / NA / NaN / N/A / − cells as unobserved, strips embedded thousand-separator commas, and skips header / metadata rows.
Reads a CSV from disk, treats blank / NA / NaN / N/A / − cells as unobserved, strips embedded thousand-separator commas, and skips header / metadata rows.
No parameter schema in public metadata yet.
to_incrementalConverts a cumulative triangle to incremental values; unobserved cells stay unobserved.
Converts a cumulative triangle to incremental values; unobserved cells stay unobserved.
No parameter schema in public metadata yet.
to_cumulativeConverts an incremental triangle to cumulative values; unobserved cells stay unobserved.
Converts an incremental triangle to cumulative values; unobserved cells stay unobserved.
No parameter schema in public metadata yet.
sample_triangleReturns the textbook 10×10 cumulative paid triangle for use as a self-check.
Returns the textbook 10×10 cumulative paid triangle for use as a self-check.
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageReadme
mcp-chainladder
Actuarial chain-ladder reserving for Claude. A Model Context Protocol server that hands Claude the tools to compute IBNR, project ultimates, run Mack (1993) stochastic error estimates, check Mack (1994) model assumptions, and parse loss triangles from CSV — all from a natural- language conversation.
Calculation only — not actuarial advice. This is a mechanical chain-ladder calculator. Selection of data, factors, tails, and exclusions, and the interpretation of stochastic error measures, require professional actuarial judgement.
What it does
Eight tools, exposed over MCP. Claude picks the right one when you ask a question; you don't have to call them by name.
Details
- Repository
- github.com/datalattice/mcp-chainladder
- Package
mcp-chainladder- Transport
- STDIO
- Registry
- MCP registry, active