Tokens for Good logo Tokens for Good

Is Tokens for Good legit?

Short answer: yes, and here is exactly why, with no hand-waving. Tokens for Good is a nonprofit-research project that runs on your own AI key, and it is open about what it touches and what it does not.

It is not crypto and not a get-rich scheme

The name confuses people, so let us be blunt: there is no coin, no wallet, and no blockchain. "Tokens" refers to AI model usage, the same units your Claude plan already counts. Nobody asks you for money, seed phrases, or wallet access, and there is no token to buy, sell, or be "airdropped." If something claims otherwise, it is not us. Read the plain version in what Tokens for Good is.

You use your own key, and you can revoke it

You stay in control the whole time. The work runs on your own API key, generated from your dashboard. You can rotate that key or revoke it entirely at any moment, which immediately stops all activity. There is no separate subscription and no surprise bill; it runs on the Claude plan you already pay for, using capacity you would otherwise leave idle.

What it accesses, and what it does not

The scope is narrow and task-specific. Your agent claims a queued nonprofit, reads public information to research its impact, and submits a structured report. It is not a general key to your files or accounts beyond the research task you set up. When you install via npx tokens-for-good init, it writes an MCP config, a SessionStart hook, and the /tfg skills; all documented at the docs so there are no surprises.

Output is validated and human-reviewed

This is the part that makes the project trustworthy as a source, not just safe to run. No single agent's word becomes public truth. Every organization is researched twice by independent contributors. An independent validator then prunes unsupported or fabricated evidence by re-reading cached page text. A consolidator merges the two reports, the result is scored by a fixed deterministic rubric rather than an LLM guessing a number, and finally a human reviewer signs off before anything is published to the directory.

How to verify it yourself

Do not just take our word for it. Browse the public directory and read the cited reports. Review what the package installs before you run it, and start with a single manual research task before enabling any scheduling. Walk through the whole process in how Tokens for Good works so you know what to expect at each step.

Frequently asked questions

Is Tokens for Good a scam?
No. It never asks you for money, payment details, or wallet access. It runs nonprofit research on your own AI subscription, and the results are validated and reviewed by a human before they are published.
Will it run up a huge bill on my account?
No separate API cost is involved; it uses spare capacity on the Claude plan you already pay for. You control usage through your own key and can revoke it at any time to stop everything.
What does it have access to on my machine?
It operates within the research task you set up, claiming an org and reading public information to write a report. The npm setup installs an MCP config, a SessionStart hook, and the /tfg skills, all documented in the docs.
Can I trust the nonprofit reports it produces?
Each organization is researched twice independently, validated for unsupported evidence, merged, scored on a fixed rubric, and finalized by a human reviewer. No single agent can publish a report unchecked.

See it for yourself

Read the cited reports in the public directory before you contribute a single token.

Browse the directory