Finlynq vs Alderfi

When to choose Alderfi

Alderfi is the right call if any of these matter more than ownership:

• You need a permissive license. Apache-2.0 has a patent grant and no copyleft, so you can wrap Alderfi in a closed-source product, ship a SaaS without sharing changes, or vendor it into a regulated stack without AGPL §13's source-disclosure trigger. Finlynq's AGPL v3 deliberately forces network forks to publish their changes — that's a feature for the maintainer and a friction for users who specifically want Apache.
• You want a local-LLM-first privacy story. Alderfi documents Llama 3 8B via llama.cpp as part of its categorization pipeline. If your threat model is "no third party — including Anthropic or OpenAI — sees my financial transactions,"Alderfi's posture matches that. Finlynq today assumes you connect a cloud LLM (Claude, ChatGPT, Cursor, Windsurf) over MCP; we don't ship a local-LLM mode.
• You want to follow a project from day one. Alderfi is openly pre-alpha and is being designed in public. If the appeal is shaping the architecture, naming the abstractions, and being one of the first contributors, an alpha project is exactly where to be.
• "The app is the MCP server" is the framing you want. Alderfi positions MCP as the core abstraction, not a layer over a traditional PFM. That's a cleaner architectural pitch than Finlynq's "PFM with first-party MCP" framing.

When to choose Finlynq

Finlynq is the right call if any of these matter:

• You want something you can use today. Finlynq is shipped: hosted at finlynq.com with a public demo (demo@finlynq.com / finlynq-demo, resets nightly) and an installable Docker + PostgreSQL stack for self-hosters. Alderfi's first alpha targets mid-May 2026 and the current public repo is a scaffold with one mock tool.
• You want a hosted option, with parity to self-host. Finlynq runs as both a managed cloud and a self-hosted Docker deployment with identical feature surfaces. Alderfi is self-host only today; the planned paid hosted tier is a separate product.
• You want a large, audited MCP tool surface today. Finlynq exposes 91 HTTP / 87 stdio tools across accounts, transactions, budgets, categories, rules, portfolio (cost basis, dividends, FX-aware aggregation), goals, loans, subscriptions, recurring transactions, FX with overrides, spending anomalies, FIRE-style cash flow forecasting, and staging-review imports. Alderfi's tool surface today is 1 mock tool.
• You want per-user encryption at rest with operator-unreadable keys. Finlynq's envelope encryption (AES-256-GCM with a scrypt-derived KEK peppered with PF_PEPPER) means even the operator running finlynq.com cannot decrypt account names, payees, notes, or tags. Alderfi's encryption-at-rest posture isn't publicly documented as of 2026-05-13.
• You need real-world import paths. Finlynq's unified staging-review pipeline ingests CSV, Excel, PDF, OFX/QFX, and email-via-Resend-Inbound into the same staged_imports table, with per-row editing, transfer-pair linking, dedup via SHA-256 over plaintext payee, and a statement-balance reconciliation callout. Alderfi documents CSV + SimpleFIN today.
• You need multi-currency. Finlynq supports 32 fiats + 4 cryptos + 4 metals, with per-currency cost-basis bucketing for portfolio holdings (a CAD account holding a USD ETF is summed in the holding's own currency, not the account's). Alderfi's multi-currency story isn't documented as of 2026-05-13.
• You want the Anthropic Connectors Directory pipeline. Finlynq submitted on 2026-05-09 and is awaiting review. Alderfi is not in the directory or pipeline as of 2026-05-13.

Side-by-side

Migrating from Alderfi

1. From Alderfi to Finlynq. Alderfi's SQLite store can be dumped via standard SQL tooling. Map transactions to CSV (date, amount, currency, payee, category, account) and feed into Finlynq's staging-review at /import/reconcile — multi-currency, transfer-pair detection, and SHA-256 dedup over plaintext payee are all built in.
2. From Finlynq to Alderfi. Finlynq's data-export endpoint produces a JSON backup with transactions, accounts, categories, portfolio holdings, and goals. You'll need to write a small JSON-to-Alderfi-import bridge once Alderfi's import surface stabilizes — neither side has a one-click migration today.
3. Hooking up your AI client. For Finlynq: Claude → Customize → Connectors → "+" → paste https://finlynq.com/mcp. OAuth 2.1 handles the rest. For Alderfi: per the repo, use npx @modelcontextprotocol/inspector npm run dev against a local server. Both projects support any MCP-compatible client.

FAQ

Sources

• alderfi.org— Alderfi project page; fetched 2026-05-13
• github.com/Earleybeast/mcp— Alderfi repo; fetched 2026-05-13
• Finlynq on GitHub— AGPL v3 source
• Finlynq MCP guide— connect Claude, Cursor, Windsurf and more