ramp-safety

Rules for safely working with Ramp, a live financial account used for company spending and payments. They cover actions that change financial records, user permissions, pagination, and money formatting.

In plain words
What is it for?
Use it whenever Ramp data is read or changed through the Ramp command-line tool or connected service, especially for approvals, payments, cards, and edits.
Why use it?
Financial actions such as approving, rejecting, paying, or editing may not be undoable. The rules reduce mistakes by requiring full details, explicit confirmation, correct amounts, and respect for the user's role.

Cursor rule

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/ramp-public/cursor-plugin/ramp-safety
Clone the repo
git clone --depth 1 https://github.com/ramp-public/cursor-plugin
Per session 571 This file is loaded in full into every session.
When invoked 571 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00571 $0.00571
Opus 5 $0.00285 $0.00285
Sonnet 5 $0.00114 $0.00114
Haiku 4.5 $0.00057 $0.00057

Measured 2d ago against content hash f56ffcc4fec1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ramp-safety scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

rules/ramp-safety.mdc · 54 lines

How it starts

The opening of the file, as written. The whole thing — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Ramp safety guardrails

These rules apply whenever you act on Ramp data — through the Ramp CLI (ramp ...) or the Ramp MCP server. They exist because Ramp tools move real money and act on a live financial account.

Write actions

  • Treat every approve, reject, payment, submit, card, or edit action as a write that cannot be undone. There is no "undo" for approvals.
  • Always show the user the full item details (amount, vendor/payee, memo, date) before approving or rejecting. Never blind-approve.
  • Confirm with the user before executing any write — especially bulk operations. State exactly what will change, then wait for explicit confirmation.
  • Rejections require a reason. Approvals don't, but a short note is helpful.
  • Respect the authenticated user's role. Most company-wide tools require admin/business-owner permissions; employees see only their own data. Surface permission errors plainly instead of retrying with different arguments.

Money formatting

Amounts are not represented consistently across resources — convert before displaying or summing:

  • Transactions: formatted strings (e.g. "$1,048.25", "-$259.49"). Strip $ and , before sorting or totaling. A leading -$ is a refund — include it but call it out.
  • Bills: numeric dollars (e.g. 15000). Use directly, no parsing.
  • Reimbursements: dollars.
  • Purchase order amounts: numeric.

There is no unified spend endpoint — card charges (transactions) and bill payments (bills) are separate. Query both for complete vendor spend. Always show a currency symbol and two decimals when presenting amounts.

Reading data

  • Paginate every list/queue until the cursor is exhausted (next_page_cursor / page_cursor) — a single page rarely returns everything.
  • Merchant/vendor names are not normalized (e.g. "Delta Air Lines" vs "Delta Airlines"). Watch for variants, dedupe by stable IDs (e.g. transaction_uuid), and flag likely merges.
  • Sort queues by highest dollar amount first when presenting them for review.

Read the full file on GitHub · 54 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 54 lines · 571 tokens per session scan A f56ffcc4fec1

Subscribe to this mod's changes

ramp-safety is a cursor rule published in the GitHub repository ramp-public/cursor-plugin (1 stars, last pushed 29d ago), licensed MIT. It adds 571 tokens to every session, about $0.0029 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.