Skip to main content
Glama
RetireGolden

RetireGolden MCP

Official
by RetireGolden

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault

No arguments

Capabilities

Features and capabilities supported by this server

CapabilityDetails
tools
{
  "listChanged": true
}
resources
{
  "listChanged": true
}

Tools

Functions exposed to the LLM to take actions

NameDescription
build_planA

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Build or replace the in-memory plan from typed household/policy params or full plan JSON.

validate_planC

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Validate plan JSON (or the current session plan).

run_projectionA

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Run a deterministic projection on the session plan. Always starts at the session plan's startYear (rebuild via build_plan to change it). detail='summary' (default) returns startYear/endYear/summary/caveats only; detail='years' also returns the full per-year ledger. Taxes are federal PLUS state and local: the resident state's modeled income tax unless assumptions.stateEffectiveTaxPct overrides it with a flat rate above 0, plus any localIncomeTaxPct. This is the same stack the RetireGolden web app runs, so for a plan exported from that app these numbers match what the user sees on screen, provided you also pass the exported startYear.

run_monte_carlo

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Run a Monte Carlo summary on the session plan. Always starts at the session plan's startYear (rebuild via build_plan to change it).

batch_evaluateB

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Evaluate many policies against the current household plan (search-friendly). Cap batches sensibly (~40 tool calls total in agent loops).

run_optimizer

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Run the engine optimizer / conversion schedule search on the session plan.

solve_max_spendingB

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Bisect maximum sustainable base annual spending for the session plan.

compare_scenariosC

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Compare two plan JSON documents via projection summaries.

explain_modeled_resultC

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Return framing, assumptions, caveats, and limitations for the current session.

get_session

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Inspect whether a plan is loaded and list caveats/conventions plus mcp/engine versions.

export_plan

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Export the current session plan as full plan JSON plus the session startYear, conventions and caveats, and the identity of the build that emitted it: schemaVersion (the engine's plan-schema version), engineVersion and mcpVersion (null if not resolvable). Round-trips via build_plan({ plan, startYear, conventions, schemaVersion, engineVersion }) — pass the exported startYear back or a non-2026 session's projection will diverge, and pass the version siblings back so a different build can warn on skew (a differing engineVersion, or a schemaVersion that disagrees with the document, is a caveat only — never a refusal). A document written against a DIFFERENT plan schema is a separate matter: the engine's validator can only read its own schema version, so such a document is rejected with an explanatory message rather than silently mis-read. Returns a clone; mutating it does not affect the live session.

describe_plan_schemaA

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Return the engine's versioned Plan JSON Schema (the source of truth for authoring a full plan document) plus its schemaVersion. Pass an optional path (dotted, e.g. 'properties.accounts.items', or JSON pointer, e.g. '/properties/accounts/items') to fetch just a subtree and keep token cost down. Read-only meta tool; also exposed as the MCP resource of the same schema. Feed a document's extracted fields into update_plan, then validate_plan.

update_plan

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Incrementally mutate the current session plan with named merge operations (add/replace/remove accounts or incomes by id; set an assumption or expense field) — for building a plan up from extracted document fragments without rebuilding each turn. Requires a seeded plan (build_plan first; NO_PLAN otherwise). The mutated plan is validated via the engine BEFORE commit: on failure the session plan is left UNCHANGED and issues are returned. Returns the updated plan summary + caveats on success.

clear_session

Educational decision-support only — not tax, legal, or financial advice. Do not prescribe securities actions. Clear the in-memory plan session.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription
plan-schemaRetireGolden engine Plan JSON Schema (v1) — the source of truth for authoring a full plan document.

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/RetireGolden/RetireGolden-MCP'

If you have feedback or need assistance with the MCP directory API, please join our Discord server