fundmyslopClaim your page

Your stuff stays yours.

The first question everyone asks: “what does this thing get access to?” Short answer: your usage numbers, and nothing else. Here’s the long answer.

0
credentials held

No OAuth, no API keys, no passwords, no cookies. Ever.

9
fields we accept

Model, four token counts, cost, duration, request + session IDs. That’s the allowlist.

1
command to undo

fundmyslop disconnect claude restores your settings exactly.

How the connection actually works

  1. 1.Device authorization. The CLI shows a short code, you approve it in your logged-in browser. The CLI receives a scoped ingestion token, prefixed fms_ingest_, that can submit telemetry and do nothing else.
  2. 2.Official telemetry, content off. The CLI backs up ~/.claude/settings.json, then merges in Claude Code’s documented OpenTelemetry variables, pointing the logs exporter at us with prompt, response, and tool-content logging explicitly disabled.
  3. 3.Server-side allowlist. Every payload passes through a parser that keeps only usage metadata. Unknown fields, content fields, and whole event types we don’t care about are dropped before anything is written. Raw payloads are never stored or logged.
  4. 4.Hashed at rest, revocable. The token is stored as a SHA-256 hash. Revoke it from the dashboard or CLI and ingestion stops immediately; your historical aggregates stay yours.

Straight answers

Does this get access to my Claude account?+

No. There is no OAuth grant, no API key, no password, no cookie. The connector configures Claude Code’s own telemetry exporter to send usage metadata to us. We could not log into your account or make requests as you even if we wanted to, because we never hold anything that could.

Can you read my prompts or code?+

No. The connector pins every Claude Code content-logging option off (prompts, responses, tool details, raw API bodies). On top of that, our server keeps a strict allowlist: model name, token counts, estimated cost, duration, and request IDs. Anything else in the payload is discarded on arrival and never stored.

What exactly does the connector change on my machine?+

It merges a handful of official OpenTelemetry environment variables into ~/.claude/settings.json after making a timestamped backup. It owns only those keys and never touches your hooks, MCP servers, permissions, or anything else. Disconnecting restores your previous values exactly.

What can the ingestion token do?+

One thing: submit usage telemetry for your connection. It cannot read your data, change settings, see your dashboard, or touch money. It is stored on our side only as a SHA-256 hash, and you can revoke it any time from the dashboard or with one CLI command.

How do I undo everything?+

Run `fundmyslop disconnect claude`. It revokes the token server-side and restores your Claude settings from the values it saved before connecting. Ingestion stops immediately. You can also revoke from the dashboard.

Why not just read my local usage logs like other trackers?+

Local trackers like CodexBar and ccusage are great for personal monitoring, and reading local files is a fine model for that. Fund My Slop moves money based on the numbers, so we use provider telemetry pushed from your machine with request IDs we can deduplicate and audit. Same privacy posture, stronger integrity.

What about payments data?+

Stripe handles all payment details. We never see card numbers or your bank account. Supporter emails are used for receipts and kept private. Payouts go directly from Stripe to your connected Stripe account.

About the dollar numbers

Claude reports an estimated, API-equivalent cost per request. On subscription plans (Pro, Max) that is not your actual bill, so we label it “estimated AI value” everywhere. Contributions are support for a person, not purchases of tokens or API credits, and the ledger only ever counts the amount actually credited after fees.

Okay, I’m in

We count the slop. We don’t read it.