[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-82895":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":14,"forks30d":14,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":9,"pushedAt":9,"updatedAt":36,"readmeContent":37,"aiSummary":38,"trendingCount":14,"starSnapshotCount":14,"syncStatus":39,"lastSyncTime":40,"discoverSource":41},82895,"repository-harness","hoangnb24\u002Frepository-harness","hoangnb24","Turn any repo into an agent-ready workspace for Claude Code, Codex, Cursor, and other coding agents.",null,"Rust",602,256,8,0,37,100,115,111,11.23,"MIT License",false,"main",[24,25,26,27,28,29,30,31,32,33,34,35],"agents-md","ai-agents","ai-assisted-development","claude-code","codex","coding-agents","context-engineering","cursor","harness-engineering","software-engineering","templates","vibe-coding","2026-06-12 02:04:29","# repository-harness\n\nTurn any software repo into an agent-ready workspace.\n\n`repository-harness` is a repository-level operating harness for Claude Code,\nCodex, Cursor, and other coding agents. It gives agents the missing project\ncontext they need before they change code: where to start, what the product\ncontract says, how risky the work is, what proof is required, and which\ndecisions future agents should inherit.\n\nThe app is what users touch. The harness is what agents touch.\n\n## Why Star This Repo\n\nStar this repo if you want practical, reusable patterns for making AI-assisted\nsoftware development more reliable, inspectable, and easier for humans to steer.\n\nThis project is exploring a simple idea:\n\n> Coding agents do not only need better prompts. They need better repositories.\n\n## The Problem\n\nMost repos are built for humans reading code in a familiar codebase. Coding\nagents usually enter with only a chat prompt and a shallow snapshot of files.\nThat leads to common failure modes:\n\n- The agent edits code before understanding product intent.\n- Important constraints live only in chat history or in someone's head.\n- Validation expectations are vague or discovered too late.\n- Architecture tradeoffs are repeated instead of inherited.\n- Large requests do not get broken into reviewable story-sized work.\n\n## The Harness Approach\n\nA repository starts to have a harness when it helps an agent answer practical\nengineering questions without relying only on chat history:\n\n- What should I read first?\n- What type of work is this?\n- Which product contract does it affect?\n- How risky is the change?\n- What proof will show the work is done?\n- What decision or lesson should future agents inherit?\n\nIn this repo, those answers live in:\n\n- `AGENTS.md` — the stable agent shim with local project notes and Harness\n  doc links.\n- `docs\u002FHARNESS.md` — the human-agent collaboration model.\n- `docs\u002FFEATURE_INTAKE.md` — tiny, normal, and high-risk work classification.\n- `docs\u002FARCHITECTURE.md` — architecture discovery and boundary rules.\n- `docs\u002FTEST_MATRIX.md` — behavior-to-proof validation expectations.\n- `docs\u002Fstories\u002F` — story packets and backlog items.\n- `docs\u002Fdecisions\u002F` — durable decisions and tradeoffs.\n- `docs\u002Ftemplates\u002F` — reusable spec, story, decision, and validation templates.\n\nOpenAI describes this shift as an agent-first world where humans steer and\nagents execute:\n\nhttps:\u002F\u002Fopenai.com\u002Findex\u002Fharness-engineering\u002F\n\n## Install Harness Into A Project\n\nFrom a target project directory, run:\n\n```bash\ncurl -fsSL \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.sh?$(date +%s)\" | bash -s -- --yes\n```\n\nOn Windows PowerShell, run:\n\n```powershell\n& ([scriptblock]::Create((irm \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.ps1\"))) -Yes\n```\n\nIf the target already has `AGENTS.md`, `docs\u002F`, or `scripts\u002F`, choose one:\n\n```bash\n# Update an existing Harness repo without moving existing files\ncurl -fsSL \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.sh?$(date +%s)\" | bash -s -- --merge --yes\n\n# Back up and replace AGENTS.md, docs\u002F, and scripts\u002F\ncurl -fsSL \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.sh?$(date +%s)\" | bash -s -- --override --yes\n```\n\n```powershell\n# Update an existing Harness repo without moving existing files\n& ([scriptblock]::Create((irm \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.ps1\"))) -Merge -Yes\n\n# Back up and replace AGENTS.md, docs\u002F, and scripts\u002F\n& ([scriptblock]::Create((irm \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.ps1\"))) -Override -Yes\n```\n\nUse `--merge` when a project already has Harness and you want to append newly\nadded Harness files without moving the existing `AGENTS.md`, `docs\u002F`, or\n`scripts\u002F` paths into backup. Existing files stay untouched; only missing\nHarness files are created.\n\nFor older Harness installs whose `AGENTS.md` still contains the full generated\noperating guide, refresh it into the small stable shim:\n\n```bash\ncurl -fsSL \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.sh?$(date +%s)\" | bash -s -- --merge --refresh-agent-shim --yes\n```\n\nThe refresh backs up the existing file. If it detects the old\nHarness-generated guide, it replaces it with the shim. If the file appears\ncustom, it appends or updates a marked Harness block instead of overwriting the\nproject's local instructions.\n\nOr install into a specific path:\n\n```bash\ncurl -fsSL \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.sh?$(date +%s)\" | bash -s -- --directory \u002Fpath\u002Fto\u002Fproject --yes\n```\n\n```powershell\n& ([scriptblock]::Create((irm \"https:\u002F\u002Fraw.githubusercontent.com\u002Fhoangnb24\u002Frepository-harness\u002Fmain\u002Fscripts\u002Finstall-harness.ps1\"))) -Directory C:\\path\\to\\project -Yes\n```\n\nUse `--dry-run` on Bash or `-DryRun` on PowerShell to preview changes before\nwriting files.\n\nThe installer also downloads the prebuilt Harness CLI for the current platform,\nverifies its `.sha256` checksum, and installs it at\n`scripts\u002Fbin\u002Fharness-cli` on macOS\u002FLinux or `scripts\u002Fbin\u002Fharness-cli.exe` on\nWindows. The Rust CLI is the main Harness tool and stable command path.\n\nHarness CLI release assets are published from tags by the\n`Harness CLI Release` GitHub Actions workflow. The installer expects each\nrelease to include `harness-cli-\u003Cplatform>` and\n`harness-cli-\u003Cplatform>.sha256` assets for macOS arm64, macOS x64, Linux x64,\nLinux arm64, and Windows x64. The Windows asset is\n`harness-cli-windows-x64.exe` plus `harness-cli-windows-x64.exe.sha256`.\n\n## Try The Flow\n\nThe fastest way to understand the harness is to inspect the tiny demo:\n\n- `docs\u002Fdemo\u002FREADME.md`: shows how a simple product idea becomes product docs,\n  stories, validation expectations, and decisions before implementation starts.\n\nA typical flow looks like this:\n\n```text\nhuman intent or product spec\n  -> product contract\n  -> feature intake\n  -> story packet\n  -> validation expectations\n  -> implementation work\n  -> decision or lesson captured for future agents\n```\n\nImplementation prompts do not go straight to code. They first pass through\nfeature intake, become story-sized work when needed, and then carry both product\nvalidation and harness maintenance expectations.\n\n## Current State\n\nThis repository is in Harness v0.\n\nThere is no application implementation and no baked-in product specification\nyet. The current work is the reusable project harness: the file structure,\nagent operating model, feature intake process, story templates, and validation\nexpectations that help humans and agents turn a future user-provided spec into\nimplementation work.\n\n## Product Sources\n\nNo product contract is currently defined.\n\nWhen a user provides a project specification, add or reference it as the input\nspec for the first buildout, then derive smaller living artifacts from it:\n\n- `docs\u002Fproduct\u002F`: current product contract files, created from the spec.\n- `docs\u002Fstories\u002F`: story packets and backlog created from selected work.\n- `docs\u002FTEST_MATRIX.md`: behavior-to-proof control panel.\n- `docs\u002Fdecisions\u002F`: durable decisions and tradeoffs.\n\nDo not keep a project-specific spec or product breakdown in this harness until\na real project supplies one.\n\n## Repository Structure\n\n```text\nproject\u002F\n  AGENTS.md\n  README.md\n  docs\u002F\n    HARNESS.md\n    FEATURE_INTAKE.md\n    ARCHITECTURE.md\n    TEST_MATRIX.md\n    HARNESS_BACKLOG.md\n    product\u002F\n    stories\u002F\n    decisions\u002F\n    demo\u002F\n    templates\u002F\n  scripts\u002F\n    README.md\n```\n\n## Contributing\n\nThis project is early and benefits most from real-world agent failure cases,\nexample harness installs, docs improvements, and reusable workflow patterns.\nSee `CONTRIBUTING.md` for contribution ideas.\n\nUseful contributions include:\n\n- Show how the harness works in a real project.\n- Add missing templates or improve existing ones.\n- Propose validation patterns for different stacks.\n- Share failures where an agent made the wrong change because the repo lacked\n  context.\n- Compare harness behavior across Claude Code, Codex, Cursor, and other tools.\n\n## Share\n\nIf this idea resonates, please star the repo and share it with someone building\nwith coding agents.\n\nShort description:\n\n> An agent-ready repo harness for Claude Code, Codex, Cursor, and other coding\n> agents: AGENTS.md, product contracts, story packets, validation matrix, and\n> decision records.\n","repository-harness 是一个将任意代码仓库转化为适用于Claude Code、Codex、Cursor等编码代理的工作空间的工具。它通过提供项目上下文，如产品需求、风险评估、验证要求和决策继承等信息，使编码代理能够更有效地理解和修改代码。该工具使用Rust语言编写，并采用MIT许可证发布。适合于希望提高AI辅助软件开发可靠性、可审查性和人类可控性的场景。通过在仓库中添加特定文档（如AGENTS.md、HARNESS.md等），帮助编码代理更好地理解项目背景及工程实践问题，从而促进人机协作效率。",2,"2026-06-11 04:09:33","CREATED_QUERY"]