project

Project rules for InferenceX, a backend service that runs and routes requests to language models. They define code boundaries, API design, configuration, and the expected development order.

In plain words
What is it for?
Adding or changing InferenceX APIs, services, inference engines, model routing, schemas, logging, and monitoring.
Why use it?
They help an agent preserve existing interfaces and keep new work in the correct layer. This reduces accidental rewrites and keeps request handling, business logic, and model routing organized.

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/coeusyk/inference-x/project
Clone the repo
git clone --depth 1 https://github.com/coeusyk/inference-x

Made for: Cursor.

Per session 653 This file is loaded in full into every session.
When invoked 653 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 $0.00653 $0.00653
Opus 5 $0.00327 $0.00327
Sonnet 5 $0.00131 $0.00131
Haiku 4.5 $0.00065 $0.00065

Measured yesterday against content hash bd05881887c6, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

project 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/project.mdc · 53 lines

How it starts

The opening of the file, as written. The whole thing — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.

InferenceX project rules

  • Treat InferenceX as a backend-first, layered LLM inference platform.
  • Preserve stable public API contracts once introduced.
  • Add new capabilities by adding modules, not by rewriting unrelated files.
  • Keep route handlers thin; put business logic in services.
  • Use typed schemas for requests, responses, and internal DTOs.
  • Prefer dependency injection over direct global imports.
  • Make behavior config-driven where possible, especially for models, routing, and logging.
  • Do not begin UI-heavy work before backend phase goals are complete.
  • When architecture changes, update docs/ARCHITECTURE.md, docs/PHASES.md, and any affected OpenSpec change docs in the same task.
  • Favor incremental edits that preserve previous working behavior.

Repo boundaries

  • src/inference_x/api/ exposes HTTP endpoints and request validation.
  • src/inference_x/services/ orchestrates application logic.
  • src/inference_x/engines/ contains inference engine interfaces and implementations.
  • src/inference_x/routing/ contains model selection logic and policies.
  • src/inference_x/observability/ contains middleware, metrics, and exporters.
  • src/inference_x/schemas/ contains Pydantic models only.
  • config/ stores YAML configuration files; avoid hardcoding deploy-specific values.

Path Handling

  • Never use absolute paths anywhere in generated code, config files, scripts, or documentation.
  • This project runs on different machines (Windows WSL2, Linux) where home directories and project roots differ. An absolute path that works on one machine will silently break on another.
  • Always use paths relative to the project root or the file being edited.
  • For Python code, resolve paths at runtime using Path(__file__).parent or Path(__file__).resolve().parent anchored to a known file in the module tree.
  • For config files (YAML, TOML), use relative paths from the config file's own location, or use a placeholder like ./models/ that the application resolves at startup.
  • For shell scripts and the Makefile, use $(shell pwd) or $(CURDIR) — never hardcode a path like /home/user/ or /mnt/c/.
  • If a path must be configurable (e.g., model weights directory), expose it as an environment variable with a sensible relative default, not a hardcoded absolute path.

Read the full file on GitHub · 53 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. yesterday First seen · 53 lines · 653 tokens per session scan A bd05881887c6

Subscribe to this mod's changes

project is a cursor rule published in the GitHub repository coeusyk/inference-x (2 stars, last pushed 21d ago), licensed MIT. It adds 653 tokens to every session, about $0.0033 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.