guidelines

General architecture rules for a modular system built from independent concepts, their specifications, and synchronizations. A specification describes the expected actions and data shapes.

In plain words
What is it for?
Use them to create missing concept implementations, maintain tests, keep concepts from importing one another, and connect synchronizations through action signatures.
Why use it?
They keep concepts independently testable and let implementations or synchronizations be regenerated from clear specifications.

Cursor rule for Cursor

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/eagonmeng/sync-blank/guidelines
Clone the repo
git clone --depth 1 https://github.com/eagonmeng/sync-blank

Made for: Cursor.

Per session 592 This file is loaded in full into every session.
When invoked 592 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 $0.00592 $0.00592
Opus 5 $0.00296 $0.00296
Sonnet 5 $0.00118 $0.00118
Haiku 4.5 $0.00059 $0.00059

Measured yesterday against content hash 562e4afbe569, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

guidelines 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.

.cursor/rules/guidelines.mdc · 54 lines

What it actually says

General:

  • You are encouraged to work with the application in phases, as each piece is modular: concept specs contain sufficient context to (re)generate independent implementations, and synchronizations can be generated without knowledge of concept implementation details, and refer solely to action signatures in their specification.
  • You should not need to inspect inside the engine/ folder unless specifically to debug a tricky issue or if you generate behavior that does not align with expectations according to your understanding of the concept + synchronization structure, and can prove so with tests.

Concept implementations should:

  • Never import one another, and should run fully independently.
  • Be able to be tested individually, and carry a canonical test modeled after their "operational principle".
  • Maintain a 1-1 correspondence with their specifications located in specs/. If there are concept specs found without a corresponding implementation, you should first generate and test an implementation.

Synchronizations should:

  • Be written concisely, with direct reference to the concept action, such as User.register as the first argument in the actions pattern lists.
  • Typing should work after concepts are instrumented, and the baseline concept User is an instance that has been instrumented.
  • Use .query on frames: Frames in the where clause function as much as possible to maintain readibility and clarity. When new bindings are provided by query, the typing should update to the next chained method's version of Frames.
  • Only query functions can be used with .query, meaning that they start with an underscore _.
  • For usage of .query, remember that the third parameter, the output pattern, will almost always expect to have a variable (deconstructed from Vars) symbol object as the value. This allows the query to hydrate the frame with that key, and types will work if you respect this pattern.
  • All indexing on frame objects in the where should only index by a var declared in the synchronization function inputs, and thus look something like $[user], and not $.user since that is a string key.
  • Use standard Array functions over manual for iteration, and maintain a functional-style approach for readibility.
  • when patterns are currently flat, meaning that you should not pattern match with nested objects, and instead flatten out parameters and bind them with the vars available.
  • This is also true for then patterns, meaning that you should transform, format, and bind any desired output in the where clause to the appropriate variable on the frame, and have simple usage of these declared variables in the output parameters.
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. yesterday First seen · 54 lines · 592 tokens per session scan A 562e4afbe569

Subscribe to this mod's changes

guidelines is a cursor rule published in the GitHub repository eagonmeng/sync-blank (4 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 592 tokens to every session, about $0.0030 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.