Provides one system's data and actions. Reads and writes inside a single product of record; it can act, but only within its own walls. More on System servers →All seven roles →
FreshBooks runs locally over stdio. You need Node.js (for npx) or Python (for uvx) installed, then add the command below to your client's MCP config file.
Create a key in the vendor's dashboard and store it as an environment variable in your client config. Never paste it into a chat window.
Step 3
Verify, then widen access
Restart your client, confirm the server's tools appear, and run one read-only question first. Keep client-side approval prompts on for any tool that can write, trade or send. Confirm with your firm's compliance policy before connecting systems holding client data.
Read-only: review my P&L and balance sheet for <period> from the accounting system and reconcile them against the bank and account-aggregation feeds for the same period. List items a buyer would question, personal expenses run through the business, uncategorised transactions, owner compensation treatment, one-off items, plus any transaction visible in the bank feed with no matching entry in the books. Output Item / Amount / Where it appears / Why a buyer asks. Do not post, edit or reclassify anything.
Guardrail · Read-only. Any adjustment goes through your accountant, not a chat tool.
Build a diligence readiness checklist across client, financial, custody and compliance records. For each item say whether it exists today, where it lives, and what work is needed. Use read-only access, keep client identities out of the output, and do not create or share a data room or upload any document.
Guardrail · Read-only. Never let a tool export client data to an external destination.
Read-only, in three passes. First gather my practice's own figures: households, AUM, recurring vs. one-time revenue, fee mix, client age distribution and top-10 concentration from CRM and custody, and revenue, expenses, owner compensation and margin from the accounting system, record where each figure came from and mark anything you could not source rather than estimating it. Second, run those inputs through ValuationMCP's guided valuation workflow, keeping the source note attached to every input you pass in and showing the methods, the range and the assumptions behind it. Third, pull the filed peer distribution for my state and AUM band from ADVmcp and use it to sanity-check the gathered inputs, client counts, account counts, custody arrangement, disclosure history, and to place my result in context. Output the valuation range, the inputs table with sources, and where the practice sits against filed peers. No client names or account numbers anywhere in the output, and do not write, export or share anything.
Guardrail · Read-only, no client identities in the output. The valuation is a modeled estimate, not an appraisal, and peer percentiles are positions rather than grades: Form ADV carries no revenue, EBITDA or margin figures, so peer data contextualises the number and never moves it.
