[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-10947":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":8,"languages":8,"totalLinesOfCode":8,"stars":9,"forks":10,"watchers":11,"openIssues":12,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":14,"stars7d":14,"stars30d":14,"stars90d":13,"forks30d":13,"starsTrendScore":11,"compositeScore":15,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":16,"fork":16,"defaultBranch":17,"hasWiki":18,"hasPages":16,"topics":19,"createdAt":8,"pushedAt":8,"updatedAt":20,"readmeContent":21,"aiSummary":22,"trendingCount":13,"starSnapshotCount":13,"syncStatus":14,"lastSyncTime":23,"discoverSource":24},10947,"hackerrank-orchestrate-may26","interviewstreet\u002Fhackerrank-orchestrate-may26","interviewstreet",null,118,384,6,4,0,2,7.76,false,"main",true,[],"2026-06-12 02:02:28","# HackerRank Orchestrate\n\nStarter repository for the **HackerRank Orchestrate** 24-hour hackathon (May 1–2, 2026).\n\nBuild a terminal-based AI agent that triages real support tickets across three product ecosystems; **HackerRank**, **Claude**, and **Visa** — using only the support corpus shipped in this repo.\n\nRead [`problem_statement.md`](.\u002Fproblem_statement.md) for the full task spec, input\u002Foutput schema, and allowed values, and [`evalutation_criteria.md`](.\u002Fevalutation_criteria.md) for how submissions are scored.\n\n---\n\n## Contents\n\n1. [Repository layout](#repository-layout)\n2. [What you need to build](#what-you-need-to-build)\n3. [Where your code goes](#where-your-code-goes)\n4. [Quickstart](#quickstart)\n5. [Chat transcript logging](#chat-transcript-logging)\n6. [Submission](#submission)\n7. [Judge interview](#judge-interview)\n8. [Evaluation criteria](#evaluation-criteria)\n\n---\n\n## Repository layout\n\n```\n.\n├── AGENTS.md                       # Rules for AI coding tools + transcript logging\n├── problem_statement.md            # Full task description and I\u002FO schema\n├── README.md                       # You are here\n├── code\u002F                           # ← Build your agent here\n│   └── main.py                     #   Entry point (rename\u002Fextend as you like)\n├── data\u002F                           # Local-only support corpus (no network needed)\n│   ├── hackerrank\u002F                 #   HackerRank help center\n│   ├── claude\u002F                     #   Claude Help Center export\n│   └── visa\u002F                       #   Visa consumer + small-business support\n└── support_tickets\u002F\n    ├── sample_support_tickets.csv  # Inputs + expected outputs (for development)\n    ├── support_tickets.csv         # Inputs only (run your agent on these)\n    └── output.csv                  # Write your agent's predictions here\n```\n\n---\n\n## What you need to build\n\nA terminal-based agent that, for each row in `support_tickets\u002Fsupport_tickets.csv`, produces:\n\n| Column         | Allowed values                                          |\n| -------------- | ------------------------------------------------------- |\n| `status`       | `replied`, `escalated`                                  |\n| `product_area` | most relevant support category \u002F domain area            |\n| `response`     | user-facing answer grounded in the provided corpus      |\n| `justification`| concise explanation of the routing\u002Fanswering decision   |\n| `request_type` | `product_issue`, `feature_request`, `bug`, `invalid`    |\n\nHard requirements (from `problem_statement.md`):\n\n- Must be **terminal-based**.\n- Must use **only the provided support corpus** (no live web calls for ground-truth answers).\n- Must **escalate** high-risk, sensitive, or unsupported cases instead of guessing.\n- Must avoid hallucinated policies or unsupported claims.\n\nBeyond that you are free to bring your own approach — RAG, vector DBs, tool use, structured output, agent frameworks, classical ML, or anything else.\n\n---\n\n## Where your code goes\n\nAll of your work belongs in [`code\u002F`](.\u002Fcode\u002F). The repo ships with an empty `code\u002Fmain.py` you can grow into your full agent — add more modules (`agent.py`, `retriever.py`, `classifier.py`, etc.) next to it as needed.\n\nConventions:\n\n- Put a **README inside `code\u002F`** describing how to install dependencies and run your agent.\n- Read secrets **from environment variables only** (`OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, …). Copy `.env.example` → `.env` (already gitignored) if you keep one. **Never hardcode keys.**\n- Be **deterministic** where possible. Seed any random sampling.\n- Write responses to `support_tickets\u002Foutput.csv`.\n\n---\n\n## Quickstart\n\nClone this repository:\n\n```bash\ngit clone git@github.com:interviewstreet\u002Fhackerrank-orchestrate-may26.git\ncd hackerrank-orchestrate-may26\n```\n\nYou are free to use any language or runtime. We recommend **Python**, **JavaScript**, or **TypeScript**.\n\n---\n\n## Chat transcript logging\n\nThis repo ships with an `AGENTS.md` that any modern AI coding tool (Cursor, Claude Code, Codex, Gemini CLI, Copilot, etc.) will read. It instructs the tool to append every conversation turn to a single shared log file:\n\n| Platform       | Path                                              |\n| -------------- | ------------------------------------------------- |\n| macOS \u002F Linux  | `$HOME\u002Fhackerrank_orchestrate\u002Flog.txt`            |\n| Windows        | `%USERPROFILE%\\hackerrank_orchestrate\\log.txt`    |\n\nYou don't need to do anything to enable it — just use your AI tool normally. You'll upload this `log.txt` as your chat transcript at submission time.\n\n---\n\n## Submission\n\nSubmit on the HackerRank Community Platform:\n\u003Chttps:\u002F\u002Fwww.hackerrank.com\u002Fcontests\u002Fhackerrank-orchestrate-may26\u002Fchallenges\u002Fsupport-agent\u002Fsubmission>\n\nYou will upload **three** files:\n\n1. **Code zip** — zip your `code\u002F` directory and upload it. Exclude virtualenvs, `node_modules`, build artifacts, the `data\u002F` corpus, and the `support_tickets\u002F` CSVs.\n2. **Predictions CSV** — your agent's output for `support_tickets\u002Fsupport_tickets.csv` (i.e. the populated `output.csv`).\n3. **Chat transcript** — the `log.txt` from the path in [Chat transcript logging](#chat-transcript-logging).\n\n---\n\n## Judge interview\n\nAfter a successful submission, your AI Judge interview will happen within a few hours after the hackathon ends. It will stay open for the next 4 hours. \n\nThe AI Judge will have access to your submission and may ask about your approach, decisions, and how you used AI while building your solution. The interview will be 30 minutes long, and keeping your camera on is mandatory.\n\nResults will be announced on May 15, 2026\n\n---\n\n## Evaluation criteria\n\nSubmissions are scored across four dimensions: agent design (your `code\u002F`), the AI Judge interview, output accuracy on `support_tickets\u002Foutput.csv`, and AI fluency from your chat transcript.\n\nSee [`evalutation_criteria.md`](.\u002Fevalutation_criteria.md) for the full rubric.","HackerRank Orchestrate 是一个为24小时黑客马拉松准备的启动项目，旨在构建基于终端的AI代理，以处理来自HackerRank、Claude和Visa三个产品生态系统的实际支持票据。核心功能包括根据提供的支持文档自动分类票据状态（如已回复或已升级）、确定相关的产品领域、生成面向用户的响应以及提供决策理由。技术上要求参与者开发的解决方案必须基于终端运行，并且仅能利用随项目提供的本地数据集，不允许进行网络调用获取额外信息。此外，对于高风险或敏感问题，系统应采取升级处理而非直接回答。该项目适用于希望探索自然语言处理与机器学习在客户服务自动化领域应用的开发者或团队。","2026-06-11 03:30:54","CREATED_QUERY"]