Firefly CFO Trust Ladder
No CFO trusts AI with company books on day one. Firefly III plus Claude is the trust ladder. Personal books first. Real reps. Then the company.
Who this is for
Best for CFOs, controllers, FP&A leaders and finance directors at $5M to $500M companies who want to actually understand AI on the books before recommending it internally.
We would deliver the full graduation playbook: self-hosted Firefly stack, Claude prompt library for the 10 trust rungs, weekly Sunday brief automation, and the personal-to-company crosswalk so the same workflow lifts straight into corporate FP&A.
Setup steps
Spin up Firefly III via Docker on a Raspberry Pi, Mac mini, or $5 a month VPS.
Import 12 months of bank statements via CSV and generate the Firefly API key.
Wire Claude to Firefly with a simple Python script or n8n flow.
Run the trust ladder rungs in order: categorise, subscription audit, cash forecast, variance memo, Sunday brief.
Before you connect live data
- • Run dummy data first. Real client data is not a test bed.
- • API keys never go in a public repo. Use env vars and a secrets manager.
- • Add logging, access control, monitoring, and a rollback path before launch.
- • Read the license. Forking a repo without checking is how lawsuits start.
Credit the original author
Firefly III is AGPL-3.0. Respect the license, especially if you modify and redistribute.
We list this as a guide, not as our build, unless we are actively maintaining a fork.
Paste this into Claude. It will install it for you.
Open Claude or Claude Code. Paste the prompt below. It will ask you a few setup questions, then walk you through every step adapted to your machine.
You are helping me install and run **Firefly CFO Trust Ladder** on my machine. Repo / link: https://github.com/firefly-iii/firefly-iii Category: Finance and data What it does: No CFO trusts AI with company books on day one. Firefly III plus Claude is the trust ladder. Personal books first. Real reps. Then the company. Difficulty: Medium Setup steps from the public guide: 1. Spin up Firefly III via Docker on a Raspberry Pi, Mac mini, or $5 a month VPS. 2. Import 12 months of bank statements via CSV and generate the Firefly API key. 3. Wire Claude to Firefly with a simple Python script or n8n flow. 4. Run the trust ladder rungs in order: categorise, subscription audit, cash forecast, variance memo, Sunday brief. Before writing or running anything, ask me these in ONE batch and wait for my reply: 1. What operating system am I on? (macOS / Windows / Linux) 2. Do I already have Docker installed and running? 3. Do I have Git installed? 4. Do I have Node or Python installed, and what versions? 5. Which model provider keys do I have available right now? (Anthropic, OpenAI, Gemini, OpenRouter, none) 6. Where do I want this installed on disk? 7. Is this just to test locally, or am I planning to connect it to live business data soon? 8. Anything sensitive (real client data, payment access, mailbox access) it should NOT touch on this machine? After I answer, walk me through each setup step adapted to my environment. Stop after each step, show me the command, and wait for me to confirm it ran before moving on. If a step fails, diagnose the actual error before suggesting the next move. Do not connect this to live business data until I explicitly say go.
Want it wired into your business instead of your laptop?
A repo on your machine is a starting point. The work that pays back is connecting it to the CRM, inbox, payments, and team processes you already run. That is the part we ship.