acb-tax-mcp
An MCP server that computes Canadian adjusted cost base (ACB) and capital gains from trade history, including average-cost tracking and superficial-loss detection, returning structured JSON.
README
acb-tax-mcp
An MCP server that computes Canadian adjusted cost base (ACB) and capital gains from your trade history: average-cost tracking, per-disposition gains, and superficial-loss detection, returned as structured JSON.
Ask your assistant "what are my capital gains for 2024?" or "did I trigger any superficial losses?" and it runs the CRA rules over your transactions instead of you wrestling a spreadsheet.
⚠️ This is a calculation aid, not tax advice. Verify every number before you file, and consult a professional for anything non-trivial. See Limitations.
Works with Claude Desktop, Claude Code, Cursor, or any MCP-compatible client.
Features
| Tool | What it does |
|---|---|
calculate_acb |
Full calculation: current holdings (shares, total ACB, ACB per share), every disposition with proceeds/ACB/outlays/gain, per-year summaries, and warnings. |
acb_summary |
Just current holdings and their book cost (handy for unrealized gains against a market price). |
capital_gains_report |
A Schedule-3-style report for one tax year: each disposition plus totals, net capital gain, and taxable gain (50% inclusion). |
check_superficial_losses |
Flags losses caught by the 30-day rule, with the denied (deferred) amount per event. |
Implements the CRA average-cost method (all shares of a security pool into one ACB; gains are against the average, not FIFO) and the superficial-loss rule (loss denied and deferred into the ACB of substitute shares bought within 30 days before or after the sale). Commissions and per-trade CAD FX conversion are handled.
Install
Requires Python 3.10+.
uv tool install acb-tax-mcp # or: pip install acb-tax-mcp
Run from source without installing:
git clone https://github.com/haiiibin/acb-tax-mcp
cd acb-tax-mcp
uv run acb-tax-mcp
Configure your client
Claude Desktop
In claude_desktop_config.json:
{
"mcpServers": {
"acb-tax": {
"command": "acb-tax-mcp"
}
}
}
Claude Code
claude mcp add acb-tax -- acb-tax-mcp
Transactions
Give the tools a list of transactions (or a path to a .csv / .json file).
| Field | Required | Notes |
|---|---|---|
date |
yes | YYYY-MM-DD |
action |
yes | buy or sell |
security |
yes | ticker / symbol (pooled by this key) |
shares |
yes | positive number |
price |
yes | price per share, in the trade currency |
commission |
no | trade commission (default 0) |
currency |
no | e.g. USD (default CAD) |
fx_rate |
no | trade currency to CAD, e.g. 1.35 for USD (default 1) |
note |
no | free text |
CSV uses the same column names as a header row.
Usage
- "Calculate the ACB and capital gains for the trades in
~/trades.csv." - "What's my capital-gains report for 2024?"
- "Did any of these sales trigger a superficial loss?"
- "What's my current book cost for XEQT?"
Example
// calculate_acb with:
// buy 100 XYZ @ $10, buy 100 XYZ @ $20, sell 100 XYZ @ $25
{
"holdings": [
{ "security": "XYZ", "shares": 100.0, "total_acb": 1500.0, "acb_per_share": 15.0 }
],
"dispositions": [
{ "date": "2024-03-01", "security": "XYZ", "shares_sold": 100.0,
"proceeds": 2500.0, "acb": 1500.0, "capital_gain": 1000.0,
"is_superficial_loss": false }
],
"summary": {
"by_tax_year": [
{ "tax_year": 2024, "net_capital_gain": 1000.0, "taxable_capital_gain": 500.0 }
],
"inclusion_rate": 0.5
}
}
Superficial loss example
Buy 100 @ $10, sell 100 @ $8 (a $200 loss), then rebuy 100 @ $8 nine days later:
{ "gain_before_superficial": -200.0, "superficial_loss_denied": 200.0,
"capital_gain": 0.0, "is_superficial_loss": true }
The $200 loss is denied and added to the ACB of the repurchased shares (new ACB per share becomes $10), so it is recovered on a future sale.
Limitations
Read these before relying on the output.
- Average-cost, per identical property. Feed all trades of the same security across your accounts together, since the CRA rule pools identical property at the taxpayer level. The tool pools by the
securitykey you provide. - Superficial losses use the standard least-of-three test with a single forward pass. Deeply chained or overlapping superficial losses can need case-by-case professional judgment.
- Not yet handled: return of capital, reinvested/notional distributions (ETF phantom distributions), stock splits, options, and other corporate actions. These affect ACB and are on the roadmap.
- FX must be supplied per transaction (use the transaction-date rate). The tool does not fetch exchange rates.
- Registered accounts (TFSA/RRSP) do not have capital gains; this tool is for non-registered (taxable) accounts.
- Not tax advice.
Development
uv venv
uv pip install -e ".[dev]"
uv run pytest
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.