npx agentspec-cli install 34059265-f3b8-4463-b655-94e8cd8b3895 --file-type cursorrules→ .cursor/rules/
# Mycelium — Cursor rules ## Project context Read the canonical brief before coding or planning: - **Primary:** `prompts/system/CORE_PROMPT.md` (foundational principles + operating model) - **Architecture & Current Direction:** `docs/architecture.md` Additional prompts live under `prompts/`: - `prompts/cursor/` — Cursor-specific task prompts and handoff system - `prompts/grok-build/` — Grok / build-agent prompts - `docs/` — architecture and human documentation ## Enforce from the brief - LangGraph (Python): explicit graphs, nodes, edges; Postgres checkpointer for state - Python 3.12+, Pydantic, full type hints; no god-agents - Layout: `src/agents/`, `src/graphs/`, `src/tools/`, `src/models/`, `src/persistence/`, `tests/` - LangSmith tracing; small reviewable diffs; clean architecture over hacks ## Workflow 1. Align with `prompts/system/CORE_PROMPT.md` 2. Propose a short plan for non-trivial work 3. Implement incrementally with types, tests where appropriate, and tracing hooks ## Cursor Agent Handoffs (Cheat Sheet) When the user says any of these: - "Work on the next task" - "Pick up the next task" - "Start the next prompt" **You must:** 1. Scan `prompts/cursor/next/` 2. Pick the oldest file (sort alphabetically) 3. **Move it immediately** to `prompts/cursor/in-progress/` (this claims the task) 4. Read and execute the prompt 5. Create output in `prompts/cursor/done/<task-name>/` with at least: - `prompt.md` (copy of the original) - `output.md` (clear summary of what was done + decisions) 6. **Remove only the specific file you claimed** from `prompts/cursor/in-progress/`. - Do **not** delete, move, or touch any other files present in `in-progress/`. - "Helpful" cleanup of other files will break parallel execution. - Note unexpected files in your output but leave them alone. **Scope Discipline (Critical):** If a prompt defines explicit scope boundaries, stay inside them. If you believe out-of-scope changes are necessary, stop and document the issue instead of making the changes. Full details: `prompts/WORKFLOW.md` Never work on a file while it's still in `prompts/cursor/next/`. The move to `in-progress/` is mandatory.
Cursor AI Assistant Rules
MANDATORY: Before ANY response, read the .cursorrules files completely. NO EXCEPTIONS.
by AgentHub Crawler
Cursor Rules — WindowConfigurator
Read `AGENTS.md` in this directory first. It contains the full working preferences, TDD rules, code conventions, architecture principles, documentation requirements, and skill registry. Do not start any task before reading it.
by AgentHub Crawler
Python Cursorrules Prompt File Best Practices Rules
Cursor IDE rules and conventions for AI-assisted development.
by AgentHub Crawler
Cursor — Bootloader Only
Sign in to leave a comment.
This file exists only to route you. The rules are NOT here. **Read `rules/DIGEST.md` first** — it is the short auto-generated list of non-negotiables every agent must follow. Then read AGENTS.md — it is the canonical agent entry point. It sends you to
by AgentHub Crawler