{
  "markdown": "# Engineering Discipline\n\n[한국어](README.ko.md) | [中文](README.cn.md)\n\nEngineering discipline skills for AI coding agents. Works with Claude Code, Gemini CLI, OpenCode, Codex, and Cursor.\n\n## How It Works\n\nSkills chain together to handle tasks from vague request to verified implementation:\n\n```\nUser request\n    |\nclarification ─── resolve ambiguity, explore codebase\n    |\n    |── Complexity Assessment (auto-routing)\n    |       |\n    |       |── Simple (score 5-8)\n    |       |       |\n    |       |       plan-crafting ─── create executable plan\n    |       |           |\n    |       |       run-plan ─── worker-validator execution loop\n    |       |           |\n    |       |       review-work ─── information-isolated verification\n    |       |\n    |       |── Complex (score 10-15)\n    |               |\n    |           milestone-planning ─── explore → draft → adversarial critics\n    |               |\n    |           long-run ─── multi-day orchestrator\n    |               |── M1: plan-crafting → run-plan → review-work → checkpoint\n    |               |── M2: plan-crafting → run-plan → review-work → checkpoint\n    |               |── ...\n    |\n    |── karpathy ─── implementation guardrails (before/during coding)\n    |── clean-ai-slop ─── post-generation AI code cleanup\n    |── simplify ─── post-implementation code quality pass\n    |── systematic-debugging ─── reproduce-first bug fixing\n    |── rob-pike ─── measurement-driven optimization\n```\n\nYou don't need to memorize this. Each skill activates automatically based on trigger phrases and context.\n\n## Skills\n\n### Workflow Skills (chain together)\n\n#### Clarification\n\nResolves vague requests into well-defined work scopes — explores the codebase first, then runs iterative Q&A grounded in the findings. Outputs a Context Brief with automatic complexity routing.\n\n**Triggers on:** \"I want to...\", \"I need...\", \"let's build...\", or any request where scope isn't immediately clear.\n\n#### Plan Crafting\n\nCreates executable multi-step implementation plans from a clear scope. Every step contains actual code — no placeholders allowed.\n\n**Triggers on:** \"plan this\", \"create a plan\", or after clarification completes with a Simple verdict.\n\n#### Run Plan\n\nExecutes plans using worker-validator pairs. Workers implement, validators verify independently with zero knowledge of the worker's approach.\n\n**Triggers on:** \"run the plan\", \"execute the plan\", or after plan-crafting completes.\n\n#### Review Work\n\nInformation-isolated post-execution verification. Reads only the plan document and the codebase — receives no execution logs or worker output.\n\n**Triggers on:** \"review the work\", \"verify the implementation\", or after run-plan completes.\n\n#### Simplify\n\nReviews changed code through three parallel agents (reuse, quality, efficiency), then fixes any issues found.\n\n**Triggers on:** \"simplify\", \"clean up the code\", \"review the changes\".\n\n### Long Running Execution\n\nFor complex tasks that span multiple days.\n\n#### Milestone Planning (Ultraplan)\n\nGrounds a draft milestone DAG in broad parallel exploration, then submits it to independent adversarial critics — hidden complexity, dependency & ordering, verifiability, integration risk — who attack the draft in rounds until no blocking findings remain.\n\n**Triggers on:** \"plan milestones\", \"break this into milestones\", \"ultraplan\", or after clarification with a Complex verdict (score 10-15).\n\n**Key features:**\n- Two draft skeletons (interface-first vs risk-first) to counter anchoring\n- Independent adversarial critics with information isolation (no cross-pollination)\n- Critique-until-dry revision loop with a verbatim resolution log\n- Binary-decidable success criteria enforced per milestone\n- Milestone count guard (warn >7, require approval >10)\n\n#### Long Run Harness\n\nOrchestrates multi-day execution. Each milestone passes through plan-crafting, run-plan, review-work with checkpoint/recovery. Runs autonomously — the milestone lock is the single approval gate; it escalates only on failures.\n\n**Triggers on:** \"long run\", \"start long run\", \"execute milestones\".\n\n**Key features:**\n- State persisted to disk after every phase transition (survives crashes)\n- Retry escalation: re-execute → re-plan → stop (with persistent counter)\n- Parallel milestones via worktree isolation\n- Recovery protocol for interrupted sessions\n- Rate limit handling aligned with Claude Code's built-in retry\n- Context window management for long conversations\n- Mid-execution correction and milestone addition procedures\n\n### Standalone Skills\n\n#### Karpathy Guidelines\n\nPreventive guardrails for code implementation — enforces surgical changes, assumption verification, and scope discipline before and during coding.\n\n**Triggers on:** Before implementing features, modifying code, or when generating code without reading existing context first.\n\n#### Clean AI Slop\n\nCorrective cleanup of AI-generated code. Removes LLM-specific patterns (over-commenting, unnecessary abstractions, defensive paranoia, verbose naming, filler) while preserving behavior. Regression-tests-first, single-smell-pass discipline.\n\n**Triggers on:** \"clean up\", \"deslop\", \"slop\", \"clean AI code\", or after any significant generation session.\n\n#### Rob Pike's 5 Rules\n\nA decision framework that prevents premature optimization and enforces measurement-driven development.\n\n**Triggers on:** \"optimize\", \"slow\", \"performance\", \"bottleneck\", \"speed up\", \"make faster\", \"too slow\"\n\n#### Systematic Debugging\n\nStrict debugging workflow: reproduce-first, root-cause-first, failing-test-first.\n\n**Triggers on:** Any bug, test failure, or unexpected behavior.\n\n**Reference guides:**\n- [Condition-based waiting](skills/systematic-debugging/condition-based-waiting.md) — Replace flaky timeouts with reliable condition polling\n- [Defense-in-depth validation](skills/systematic-debugging/defense-in-depth.md) — Validate at every layer data passes through\n- [Root cause tracing](skills/systematic-debugging/root-cause-tracing.md) — Trace backward through call chains to find the original trigger\n- [Find polluter script](skills/systematic-debugging/find-polluter.sh) — Bisection script to find which test creates unwanted files/state\n\n## Quick Start\n\nAfter installation, just describe what you want to do:\n\n- **\"I want to add authentication to the API\"** — triggers clarification, routes to plan-crafting or milestone-planning based on complexity\n- **\"run the plan\"** — executes the plan with worker-validator verification\n- **\"this test is flaky\"** — triggers systematic-debugging\n- **\"the API is slow\"** — triggers rob-pike measurement-first workflow\n- **\"clean up the AI code\"** — triggers clean-ai-slop corrective cleanup\n- **\"simplify\"** — reviews recent changes for reuse, quality, and efficiency\n- **\"long run\"** — starts multi-day milestone execution with checkpoints\n\n## Installation\n\n### Claude Code\n\n```\n/plugin marketplace add tmdgusya/engineering-discipline\n/plugin install engineering-discipline\n```\n\n### Gemini CLI\n\n```bash\ngemini extensions install https://github.com/tmdgusya/engineering-discipline\n```\n\n### Cursor\n\nInstall from the plugin marketplace, or:\n\n```text\n/add-plugin engineering-discipline\n```\n\n### Codex\n\n```bash\nnpx skills add tmdgusya/engineering-discipline\n```\n\nOr install globally (available across all projects):\n\n```bash\nnpx skills add tmdgusya/engineering-discipline -g\n```\n\nSee [Codex install guide](.codex/INSTALL.md) for details.\n\n### OpenCode\n\nAdd to your `opencode.json`:\n\n```json\n{\n  \"plugin\": [\"engineering-discipline@git+https://github.com/tmdgusya/engineering-discipline.git\"]\n}\n```\n\nSee [OpenCode install guide](.opencode/INSTALL.md) for details.\n\n## Verify Installation\n\nStart a new session and mention a performance concern or a bug. The relevant skill should activate automatically.\n\n## Marketplace\n\nThis plugin is listed in the Claude Code plugin marketplace.\n\n- **Category:** engineering\n- **Tags:** optimization, debugging, engineering, discipline, rob-pike, systematic, planning, long-running\n\n### Install from marketplace\n\n```\n/plugin marketplace add tmdgusya/engineering-discipline\n/plugin install engineering-discipline\n```\n\n### Publish updates\n\nUpdate `.claude-plugin/marketplace.json` with a new version and push to the repository. The marketplace entry pulls metadata from that file.\n\n## License\n\nMIT\n",
  "bytes": 8323,
  "sha": "8abeb22852ff09056825e9c0a40f58a4a528aa1c9ea71464b587ea69a5b8c2b3",
  "repo_slug": "tmdgusya/engineering-discipline",
  "fonte": "repo",
  "truncated": false,
  "api": "https://api.agentalog.com/api/listings/skl_tmdgusya_engineering_discipline_review_w_299b1cc4/readme"
}