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.
npx agentmods add rules/prisma/prisma-next/adr-writinggit clone --depth 1 https://github.com/prisma/prisma-nextWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.00402 |
| Opus 5 | $0.00000 | $0.00201 |
| Sonnet 5 | $0.00000 | $0.00080 |
| Haiku 4.5 | $0.00000 | $0.00040 |
Grade A, and why
adr-writing 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 yesterday.
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.
This is a copy
88% identical to adr-writing — 1 line differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
What it actually says
- Lead with the decision and a grounding example. Open with a representative code/usage sample and the decision itself, then build up why → how → consequences, and put alternatives considered last. Don't front-load context/problem/rejected designs — a reader should see what the decision is before the information being made irrelevant. (Context still belongs in the doc; it just doesn't lead.)
- ADRs are long-lived — no transient references. No PR numbers, and no "before this refactor / earlier implementation / as-shipped / what we used to do" preceding-state framing. The shape you replaced is a legitimate alternative considered, stated timelessly — not narrated as a project event. (Ticket / milestone / task IDs are already banned in ADRs by
no-transient-project-ids-in-code.mdc.) - Keep responsibility splits explicit (e.g. adapter lowering vs codec encode/decode vs driver transport/normalization)
- If an ADR discusses capabilities, follow
.cursor/rules/capabilities-ownership.mdcfor wording - If you need to describe sequencing (order of operations), keep it short and place it near the responsibilities/decision; avoid long procedural sections that read out-of-place
- Use fenced code blocks for SQL (
```sql) and TS (```ts); avoid nested backticks inside inline code - Worked examples should be end-to-end and make it obvious which layer owns each step (lane, adapter, codec, driver, runtime)
- If the decision instantiates a structural pattern recorded in
docs/architecture docs/patterns/, link forward to the pattern entry from the ADR (existing ADRs are not retroactively edited)
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.
- yesterday First seen · 16 lines · 0 tokens per session scan A 596945d7ae8d
adr-writing is a cursor rule published in the GitHub repository prisma/prisma-next (421 stars, last pushed 7d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 402 tokens. A static security scan graded it A with 0 findings. It is 88% identical to adr-writing, differing in 1 line, and is treated as a copy.
Other cursor rules, from other repositories
cli-error-handling
CLI command error handling patterns.
prefer-direct-imports-over-module-mocks
Prefer extracting a testable core over vi.mock / vi.resetModules when unit tests need to reach production logic entangled with config, env, or singletons.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.
directory-layout
Directory layout rules for SQL family vs concrete targets.
family-instance-domain-actions
Family instance domain action implementation patterns.
import-validation
Import validation and package layering rules.