Boldin

Reads the Roth conversions modeled in the user's plan: planned transfers from tax-deferred accounts (IRA, 401k, etc.) to Roth accounts, as entered in the plan. Use scenario_id to read any other saved scenario. Does not move money.

get_roth_conversions

Look up the reported fundamentals Boldin holds for a stock, ETF, or mutual fund — company profile and sector, valuation multiples, reported quarterly and annual earnings, income statement / balance sheet / cash flow lines, dividends, and for a fund its expense ratio, yield, sector and regional mix, and top 10 holdings. Use it when the user asks what a holding of theirs actually is, what it costs, or how the company behind it has been performing. Every figure is as reported on the date in `fundamentalsAsOf` — say that date when you quote a number, and never present this data as a reason to buy, sell, or hold anything.

get_security_fundamentals

Lists all of the user's saved scenarios. Returns scenario names, IDs, retirement dates, and metadata. Use this to discover scenarios before comparing them with compare_saved_scenarios, and to name a scenario's retirement date instead of inferring it.

list_user_scenarios

Proposes a reversal of the most recent plan change the user APPLIED via review_plan_changes and apply_plan_changes. Loads the prior values into hypothetical modifications, which the user commits via review_plan_changes and apply_plan_changes again. USE THIS whenever the user says 'undo', 'revert', 'take back', or asks to roll back a change — unless there are currently unapplied hypotheticals they're trying to abandon, in which case use clear_hypotheticals.

undo_last_plan_change

Summarizes the hypothetical modifications staged in this session as the exact changes they would make to the user's saved plan, and returns the confirmation_token that apply_plan_changes requires. ALWAYS show the user this summary and wait for their explicit go-ahead before calling apply_plan_changes - nothing is saved by this tool.

review_plan_changes

REQUIRED for ALL mathematical calculations — NEVER do arithmetic without this tool. Use this for compound interest, amortization schedules, tax bracket math, savings targets, scenario loops, or any computation the other tools don't cover. The code runs in a sandboxed JavaScript environment (5 seconds timeout). Push final results to the `_results` array so they are returned.

run_calculation

Runs a financial forecast. Returns summary metrics by default. Set use_modifications: true to include hypothetical changes from update_plan_scenario (current scenario only). Use scenario_id to run a forecast for any other saved scenario. Read get_planning_reference(topic: 'forecast') before interpreting the result: every money figure carries its own basis.

run_forecast

Runs guardrails analysis. Use scenario_id to analyze any other saved scenario. A risk-based retirement spending strategy that uses Monte Carlo simulation to determine: (1) a safe monthly spending target throughout retirement, (2) savings balance thresholds (guardrails) signaling when to adjust spending. Returns plannedMonthlySpending (current plan), targetSpending (savingsBalance + monthlyAmount), upperGuardrail, lowerGuardrail, and retirementAnchor — the month every figure is solved from, which you must read before explaining any difference between two scenarios. All values are total monthly spending including estimated taxes, healthcare, and all plan expenses, expressed in today's dollars. CUSTOM CONFIDENCE THRESHOLDS: Users may want to model guardrails with different confidence levels. Use target_cos, upper_cos, and lower_cos parameters to override the defaults. Constraints: A minimum 5% gap between adjacent thresholds is required. Specifically: lower_cos <= (target_cos - 0.05) and target_cos <= (upper_cos - 0.05). Also: lower_cos >= 0.20 and upper_cos <= 0.99. Defaults: target_cos=0.80, upper_cos=0.95, lower_cos=0.70. Example: A user says "What if I targeted 90% confidence?" → set target_cos=0.90, upper_cos=0.97, lower_cos=0.85. Always pass all three together — moving target_cos alone leaves the defaults too close to it and the call is rejected. Set use_modifications: true to include hypothetical changes from update_plan_scenario. Read get_planning_reference(topic: 'guardrails') before explaining the result.

run_guardrails

Retrieves Boldin resources to provide answers, about the platform, financial planning articles and live events. Use 'help' for how-to guides, 'blog' for retirement articles, 'events' for upcoming live webinars, 'videos' for YouTube videos, or 'all' for everything.

search_help

Search Boldin's verified securities (stocks, ETFs, funds) by ticker, name, or both ('FXNAX Fidelity U.S. Bond Index Fund') to get the exact ticker, name, type, CUSIP, and ISIN; a partial or misspelled name still matches. ALWAYS use this to confirm the correct security before adding or editing an investment holding — never guess a ticker or identifier. Then pass the returned ticker/name/type/cusip/isin into update_plan_scenario's investmentHoldings. If several securities match, ask the user which one they mean. If nothing matches, the security isn't in Boldin's reference data yet — proceed with the user's own ticker/name.

search_securities

Searches the user's uploaded financial documents (tax returns, brokerage statements, insurance policies, etc.) stored in their personal Vault. Use when the user references something they've uploaded, asks about a specific document, or wants details that would only be in their financial paperwork. Results include the user's full document list (with processing status and chunk_count), so also use this when they ask what's in their vault. A query alone returns only the closest-matching excerpts, which will miss line items when a statement holds many accounts or a table spanning several pages: pass doc_id to get that whole document back instead, and always do so before listing holdings, accounts, or any other set you need in full. A doc_id result reporting retrieval: excerpts means the document was too large to return whole, so you have only part of it. A whole statement comes with positions_listed, the positions it holds grouped by the account printed on the statement, so you can check a transcription against it before writing holdings.

search_vault

Send feedback about this Boldin MCP server to the Boldin team that builds it. Use this when: - The user needed something no tool on this server can do. - A tool returned wrong, incomplete or confusing results. - A task took many calls or workarounds that could be simpler. - A tool description or parameter was misleading. - The user asks to send feedback about this server. Report on your own initiative only when a tool here blocked the task or forced a workaround. Do NOT use this for: - Problems with other MCP servers or tools. Tell the user instead. - The Boldin app itself, billing, or account support. Nobody replies here and no support ticket is created; point the user to help.boldin.com. - Tests, placeholders, duplicates, or "no issues" reports. If there is nothing concrete to report, do not call this tool. - Transient errors that succeeded on retry. Privacy rules: - Never include names, emails, account numbers, balances, income, ages, dates of birth, or any other personal or financial details. - task_context, expected and actual must describe the goal in general terms. Good: "User wanted to compare two Social Security claiming ages." Bad: "User (62, $1.2M portfolio) wanted to claim at 67." - Only fill user_quote if the user raised the issue and explicitly agreed to share their words. Ask them first. Submit one report per distinct issue per conversation.

submit_feedback

Temporarily modifies a scenario for 'what-if' analysis. You MUST pass the scenario_id of the scenario you intend to modify. Only the current conversation scenario (is_current_conversation: true) can be modified. Call get_plan_schema first for the field names and modification rules. Changes are hypothetical and live only in this session - they are written to the user's plan only after the user reviews them (review_plan_changes) and explicitly confirms, at which point you call apply_plan_changes.

update_plan_scenario

Moves this session onto another of the user's saved scenarios: update_plan_scenario, review_plan_changes and apply_plan_changes then work on it, and get_user_plan and run_forecast default to it. Get scenario IDs from list_user_scenarios. Not allowed while changes are staged - review and apply them, or clear_hypotheticals, first.

switch_scenario