{"id":50972892,"url":"https://github.com/madebyaris/cursor-composer-rules","last_synced_at":"2026-06-19T04:02:18.809Z","repository":{"id":358894335,"uuid":"1241159106","full_name":"madebyaris/cursor-composer-rules","owner":"madebyaris","description":"How a senior engineer thinks, encoded for Cursor agents — evidence-first, surgical edits, reasoning \u0026 metacognition, and honest verification. Stack-agnostic rules, skills \u0026 subagents tuned for Composer.","archived":false,"fork":false,"pushed_at":"2026-06-15T03:51:56.000Z","size":57,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-06-15T05:15:40.364Z","etag":null,"topics":["agent-rules","ai-coding-assitance","composer","cursor","cursor-ai","cursor-composer","cursor-ide","cursor-rules","developer-tools","llm"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/madebyaris.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-05-17T03:05:48.000Z","updated_at":"2026-06-15T03:52:00.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/madebyaris/cursor-composer-rules","commit_stats":null,"previous_names":["madebyaris/cursor-composer-rules"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/madebyaris/cursor-composer-rules","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/madebyaris%2Fcursor-composer-rules","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/madebyaris%2Fcursor-composer-rules/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/madebyaris%2Fcursor-composer-rules/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/madebyaris%2Fcursor-composer-rules/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/madebyaris","download_url":"https://codeload.github.com/madebyaris/cursor-composer-rules/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/madebyaris%2Fcursor-composer-rules/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34516549,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-19T02:00:06.005Z","response_time":61,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agent-rules","ai-coding-assitance","composer","cursor","cursor-ai","cursor-composer","cursor-ide","cursor-rules","developer-tools","llm"],"created_at":"2026-06-19T04:02:18.015Z","updated_at":"2026-06-19T04:02:18.804Z","avatar_url":"https://github.com/madebyaris.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n# 🎼 Composer-era Cursor Rules\n\n**How a senior engineer thinks — encoded for Cursor agents.**\n\nEvidence over assumption · surgical edits · honest uncertainty · root-cause debugging · structured research · sane orchestration\n\n\u003cbr /\u003e\n\n[![License: MIT](https://img.shields.io/badge/License-MIT-22c55e?style=for-the-badge)](LICENSE)\n[![Stack](https://img.shields.io/badge/Stack-agnostic-6366f1?style=for-the-badge)]()\n[![Cursor](https://img.shields.io/badge/Cursor-Rules_\u0026_Skills-0ea5e9?style=for-the-badge)](https://cursor.com)\n[![Composer 2.5](https://img.shields.io/badge/Tuned_for-Composer_2.5-f97316?style=for-the-badge)]()\n\n\u003cbr /\u003e\n\n[Quick start](#-quick-start) · [What's new](#-whats-new-25-refresh) · [Rules](#-what-the-pack-teaches) · [Design](#-design-principles) · [Extend](#-extending)\n\n\u003c/div\u003e\n\n---\n\n## ✨ At a glance\n\n| | Principle | What you get |\n| --- | --- | --- |\n| 🔍 | **Evidence first** | Read code, run commands, cite paths — never guess |\n| 🤔 | **Reason, then re-check** | Infer the real ask; challenge the plan before acting; push back honestly |\n| ✂️ | **Surgical edits** | Match project style; smallest diff that proves the next step |\n| 📋 | **Rich plans** | Structured plan template with options, inventory, verification matrix |\n| 🛡️ | **Git safety** | Never `git push` unprompted — commits ≠ remote updates |\n| 🧠 | **Senior practices** | Version-pinned guidance from official docs, not stale training data |\n| ✅ | **Proof contract** | `verified` / `implemented but unverified` / `blocked` — no fake \"done\" |\n\nStack-agnostic. **Two always-on rules** form the operating spine; everything else loads on demand by description match. No model identity claims — Cursor's lineup changes; the principles don't.\n\n```text\n┌─────────────────────────────────────────────────────────────┐\n│  ALWAYS ON          │  REQUESTABLE ON TASK                  │\n│  composer-core      │  reasoning · orchestration · coding   │\n│  composer-verify    │  debugging · research · delivery      │\n│                     │  tools · clarify                      │\n├─────────────────────┼───────────────────────────────────────┤\n│  AGENTS             │  SKILLS                               │\n│  verifier · debugger│  deep-research · senior-practices     │\n└─────────────────────────────────────────────────────────────┘\n```\n\n---\n\n## 🚀 Quick start\n\nRepository: **`madebyaris/cursor-composer-rules`**\n\n```bash\ngit clone https://github.com/madebyaris/cursor-composer-rules.git\n\n# Replace the destination with your actual app directory\ncp -r cursor-composer-rules/.cursor path/to/your-app/.cursor\n```\n\n\u003e **⚠️ Merge carefully** if you already have `.cursor/rules` or `.cursor/agents`:\n\u003e\n\u003e - Duplicate **always-on** rules hurt quality — keep one core + one verification spine.\n\u003e - If you already have a `verifier` subagent, merge prompts or rename to avoid conflicting delegation.\n\u003e - Custom subagents in this pack are examples; adapt descriptions to your team.\n\n---\n\n## 🆕 What's new (2.5 refresh)\n\n| Addition | Purpose |\n| --- | --- |\n| [`composer-reasoning.mdc`](.cursor/rules/composer-reasoning.mdc) | **Senior reasoning + metacognition** — infer the real ask, reason then re-evaluate, principal judgment (tradeoffs, reversibility), honest pushback |\n| [`composer-core.mdc`](.cursor/rules/composer-core.mdc) | **Intent + re-evaluate** principles; effort calibration tier; sharper plan vs act; **git remote safety** (never push unprompted) |\n| [`composer-orchestration.mdc`](.cursor/rules/composer-orchestration.mdc) | When to delegate vs inline; **structured plan template**; foreground/background subagents; long-running checkpoints |\n| [`composer-coding-excellence.mdc`](.cursor/rules/composer-coding-excellence.mdc) | **Style governance** — match vs improve decision tree; no drive-by refactors |\n| [`composer-senior-practices.mdc`](.cursor/rules/composer-senior-practices.mdc) | Version-pinned best practices from primary docs + corroborated guidance |\n| [`cursor-tools-discipline.mdc`](.cursor/rules/cursor-tools-discipline.mdc) | Explore/Bash/Browser subagents, background work, sandbox **blocked** handling; **git command allow/deny** |\n| [`.cursor/skills/senior-practices/`](.cursor/skills/senior-practices/SKILL.md) | Workflow to anchor framework guidance to repo versions and credible sources |\n| [`.cursor/agents/verifier.md`](.cursor/agents/verifier.md) | Skeptical post-work check before **verified** |\n| [`.cursor/agents/debugger.md`](.cursor/agents/debugger.md) | Isolated root-cause pass (repro first in parent) |\n\nOptional: extend the pack with [Cursor Marketplace](https://cursor.com/marketplace) plugins (skills, MCP, hooks) — not required for the core behavior contract.\n\n---\n\n## 📚 What the pack teaches\n\n### 🔒 Always-on (the operating spine)\n\n| File | Purpose |\n| --- | --- |\n| [`composer-core.mdc`](.cursor/rules/composer-core.mdc) | Principles: evidence-first, read before edit, smallest diff, effort calibration, plan vs act |\n| [`composer-verification.mdc`](.cursor/rules/composer-verification.mdc) | Proof contract: `verified` / `implemented but unverified` / `blocked`; independent verification |\n\n### 📎 Requestable (loaded by description match)\n\n| File | Use for |\n| --- | --- |\n| [`composer-reasoning.mdc`](.cursor/rules/composer-reasoning.mdc) | Ambiguous, high-stakes, or architectural work: infer real intent, reason then re-evaluate, weigh tradeoffs, push back honestly |\n| [`composer-orchestration.mdc`](.cursor/rules/composer-orchestration.mdc) | Subagents, plan mode, parallel workstreams, long-running tasks |\n| [`composer-coding-excellence.mdc`](.cursor/rules/composer-coding-excellence.mdc) | Writing or modifying code: surgical edits, convention matching, style governance, no fabricated APIs |\n| [`composer-senior-practices.mdc`](.cursor/rules/composer-senior-practices.mdc) | Version-pinned framework/coding guidance from primary docs and corroborated sources |\n| [`composer-debugging.mdc`](.cursor/rules/composer-debugging.mdc) | Broken or unexpected behavior: reproduce, trace, root cause |\n| [`composer-fullstack-delivery.mdc`](.cursor/rules/composer-fullstack-delivery.mdc) | Multi-layer changes: seams, vertical slice, failure modes |\n| [`composer-deep-research.mdc`](.cursor/rules/composer-deep-research.mdc) | Audits, comparisons, benchmarks: primaries, claim ledger, contradictions |\n| [`cursor-tools-discipline.mdc`](.cursor/rules/cursor-tools-discipline.mdc) | Tools: parallel reads, MCP schemas, subagents, no fabricated output |\n| [`clarify-first.mdc`](.cursor/rules/clarify-first.mdc) | When to ask (after inspect); plan-mode exception |\n\n### 🤖 Example subagents\n\n| File | Use for |\n| --- | --- |\n| [`verifier.md`](.cursor/agents/verifier.md) | After \"done\" claims — run checks, report verified vs gaps (`readonly`) |\n| [`debugger.md`](.cursor/agents/debugger.md) | Isolated debug when repro is known — root cause + minimal fix |\n\nBuilt-in Explore, Bash, and Browser subagents need no config; orchestration rule explains when Cursor uses them.\n\nInvoke explicitly: `/verifier confirm tests pass`, or ask naturally: *\"use the debugger subagent on this failure.\"*\n\n### 🛠 Skills (long-form workflows)\n\n| Path | Purpose |\n| --- | --- |\n| [`.cursor/skills/deep-research/SKILL.md`](.cursor/skills/deep-research/SKILL.md) | Multi-pass research with claim ledger and contradiction log |\n| [`.cursor/skills/deep-research/reference.md`](.cursor/skills/deep-research/reference.md) | Compact checklist version |\n| [`.cursor/skills/senior-practices/SKILL.md`](.cursor/skills/senior-practices/SKILL.md) | Version-pinned best practices: inventory → T1 anchor → reconcile with repo |\n| [`.cursor/skills/senior-practices/reference.md`](.cursor/skills/senior-practices/reference.md) | Compact checklist version |\n\n---\n\n## 🧭 Design principles\n\n| | Principle |\n| --- | --- |\n| 🎯 | **Behavior contracts, not model identity.** Rules don't assert \"Composer = X model\". Cursor's model lineup changes; the principles are durable. |\n| 🤔 | **Intent + metacognition.** Read for the real outcome, not the literal words; form an approach, then re-evaluate it before acting. Depth scales with blast radius. |\n| 🪶 | **Tiny always-on core.** Always-on prompts cost context forever. Orchestration, tools, and specialists stay requestable or in `.cursor/agents/`. |\n| 🧩 | **One concern per rule.** Orchestration is separate from debugging is separate from research. Activate what the task needs. |\n| 🔧 | **Concrete over abstract.** Rules show good vs bad with examples, not adjectives. |\n| 🪞 | **Honest tooling.** Never fabricate output; never mark **verified** from a subagent summary alone. |\n| 🚫 | **Git remote safety.** Agents must not `git push` unless the user explicitly asks in the current turn. For hard enforcement, add your own [`beforeShellExecution` hook](https://cursor.com/docs/agent/hooks) — not shipped in this pack. |\n| ⚖️ | **Agents execute, rules govern.** Verifier runs checks; verification rule defines when and how to label results. |\n\n---\n\n## 🔧 Extending\n\nAdd stack-specific rules as new `.mdc` files with `globs`:\n\n```text\n.cursor/rules/typescript.mdc        # globs: **/*.{ts,tsx}\n.cursor/rules/python.mdc            # globs: **/*.py\n.cursor/rules/rust.mdc              # globs: **/*.rs\n```\n\nAdd focused subagents under `.cursor/agents/` with specific `description` fields — avoid dozens of vague helpers ([Cursor subagent docs](https://cursor.com/docs/agent/subagents)).\n\nKeep extensions short. Don't duplicate core or verification — link to them.\n\n---\n\n## 📁 Layout\n\n```text\n.cursor/\n  agents/\n    verifier.md                      # example subagent\n    debugger.md                      # example subagent\n  rules/\n    composer-core.mdc                # always on\n    composer-verification.mdc        # always on\n    composer-reasoning.mdc           # requestable\n    composer-orchestration.mdc       # requestable\n    composer-coding-excellence.mdc   # requestable\n    composer-senior-practices.mdc    # requestable\n    composer-debugging.mdc           # requestable\n    composer-deep-research.mdc       # requestable\n    composer-fullstack-delivery.mdc  # requestable\n    cursor-tools-discipline.mdc      # requestable\n    clarify-first.mdc                # requestable\n  skills/\n    deep-research/\n      SKILL.md\n      reference.md\n    senior-practices/\n      SKILL.md\n      reference.md\nREADME.md\nLICENSE\n.gitignore\n```\n\n---\n\n\u003cdiv align=\"center\"\u003e\n\n## 📄 License\n\nMIT — see [LICENSE](LICENSE)\n\n\u003cbr /\u003e\n\n**Made for teams who want agents that ship evidence, not vibes.**\n\n\u003c/div\u003e\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmadebyaris%2Fcursor-composer-rules","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmadebyaris%2Fcursor-composer-rules","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmadebyaris%2Fcursor-composer-rules/lists"}