[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80718":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":18,"rankGlobal":10,"rankLanguage":10,"license":19,"archived":20,"fork":20,"defaultBranch":21,"hasWiki":22,"hasPages":20,"topics":23,"createdAt":10,"pushedAt":10,"updatedAt":33,"readmeContent":34,"aiSummary":35,"trendingCount":15,"starSnapshotCount":15,"syncStatus":36,"lastSyncTime":37,"discoverSource":38},80718,"vowline","chojondocho\u002Fvowline","chojondocho","A compact operating covenant for AI agents.","",null,"Python",49,6,44,0,3,5,44.54,"MIT License",false,"main",true,[24,25,26,27,28,29,30,31,32],"agent-skills","ai-agents","claude-code","codex","developer-tools","gpt-5-5","prompt-engineering","skill-md","skills","2026-06-12 04:01:29","# Vowline\n\n[![check](https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Factions\u002Fworkflows\u002Fcheck.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Factions\u002Fworkflows\u002Fcheck.yml)\n[![release](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fchojondocho\u002Fvowline?label=release&cacheSeconds=300)](https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Freleases)\n[![license](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-blue.svg)](LICENSE)\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fvowline-banner.png\" alt=\"Vowline brand banner: one luminous line moving through aligned agent nodes\" width=\"100%\">\n\u003C\u002Fp>\n\n**Universal operating skill: bind the public task, act from evidence, preserve boundaries, and verify the delivered result.**\n\nVowline is a portable `SKILL.md` package for Codex, Claude Code, Windsurf, Cursor, Gemini CLI, GitHub Copilot, and other skill-compatible agent harnesses. It gives agents the compact operating contract in `skills\u002Fvowline\u002FSKILL.md`: bind the target and future client once, convert visible requirements into predicates over the final surface, preserve public labels and source-backed entity boundaries, prefer native structured tools, edit only the smallest live write set, and verify the delivered surface through the same interface a real user or official consumer will use.\n\nUse it for substantive work with a real deliverable, constraints, evidence requirements, cleanup risk, data interpretation, artifact fidelity, repository scope, or completion criteria. Vowline never overrides higher-priority system, tool, safety, project, or explicit user instructions.\n\n## Current benchmark snapshot\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fbenchmarks\u002F6057fa00_benchmark_snapshot.svg\" alt=\"Vowline 0.7.0(6057fa00) passes 12 of 13 checked deterministic benchmark tasks at gpt-5.5 xhigh\" width=\"100%\">\n\u003C\u002Fp>\n\nA deterministic benchmark snapshot for the exact Vowline 0.7.0 skill body (`6057fa00a6c074fe2af5f28b0a11062e69e154acf6296d4c65eb8b63f1cd637c`) passes **12 \u002F 13 checked tasks (92.3%)** at `gpt-5.5` with `xhigh` reasoning effort. The completed public\u002Fverifier-backed rows span SkillsBench, IFEval, LiveBench, and Terminal-Bench; `nginx-request-logging` is the only open task. This is mostly single-run deterministic verifier evidence, not a `pass@2` score or repeated-stability claim. No proxy judges, no LLM-as-judge, no mixed-hash rows, and no all-pass claim. See [`6057fa00_benchmark_snapshot.md`](docs\u002Fbenchmarks\u002F6057fa00_benchmark_snapshot.md) for the exact rows and result artifact paths. The earlier 0.6.0 failure-only comparison remains archived at [`8e427bb4_no_skill_comparison.md`](docs\u002Fbenchmarks\u002F8e427bb4_no_skill_comparison.md).\n\n**This is not a benchmark-overfit skill.** The tested Vowline contract is a general operating discipline: target and future-client binding, final-surface predicates, public-label preservation, declared-source entity boundaries, scoped live writes, native structured tools, faithful consumer proof, and explicit reporting. It forbids encoding unavailable private facts, hidden tests, private fixtures, benchmark quirks, known answers, one-off paths, measurement tricks, hardcoded outputs, or verifier workarounds.\n\n## Quick start\n\nThe easiest path is to ask the target agent to install Vowline for itself:\n\n```text\nInstall Vowline for yourself by following https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Fblob\u002Fmain\u002FINSTALL.md. Verify installation.\n```\n\nFor a project-local install, ask from inside the project:\n\n```text\nInstall Vowline into this project by following https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Fblob\u002Fmain\u002FINSTALL.md. Use project-local paths. Verify installation.\n```\n\nTo update an existing install after a Vowline release or contract change:\n\n```text\nUpdate Vowline for yourself by following https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Fblob\u002Fmain\u002FUPDATE.md. Verify the update.\n```\n\nYou do not need to clone this repository if your agent can read GitHub and write the relevant local files. Vowline is designed to be installed by agents as well as by humans.\n\n## What Vowline changes\n\nVowline does not add a rigid workflow. It changes the agent's default operating standard for work where incomplete evidence, unchecked artifacts, overbroad edits, hidden checker adaptation, malformed proof predicates, or vague completion claims would create failure.\n\n| Area | 0.7.0 operating requirement |\n| --- | --- |\n| Public contract | Bind the target, future client, invocation surface, inputs, output surface, allowed writes, exclusions, success threshold, budget, and final-response shape once. |\n| Working ledger | Track facts, sources, exceptions, references, unknowns, decisions, and proof predicates while separating declarations from references, examples from defaults, owner surfaces from fragments, and absence from zero. |\n| Final-surface predicates | Convert requirements into checks over required and forbidden bytes, names, literals, case, wrappers, starts, endings, order, counts, domains, precision, persistence, and side effects. |\n| Public labels | Preserve distinctive public modifiers and nouns when labels become internal or visible names; avoid generic carriers, synonyms, and private relabeling. |\n| Source authority | Build entity rows, mappings, joins, aggregations, metrics, and native artifacts from declared sources and consumer-visible contracts instead of weak compatibility or sample-only support. |\n| Tooling and proof | Prefer native structured tools and prove the final artifact or state through the same verifier, parser, renderer, workflow, service, or consumer that will use it. |\n| Transferability | Optimize for general competence, not unavailable private facts, known answers, one-off paths, measurement tricks, or verifier workarounds. |\n\nIn practice, Vowline keeps agents focused on the same shape as the canonical `SKILL.md`:\n\n```text\nRespect higher-priority system, tool, safety, project, and explicit user instructions.\nBind the target, future client, invocation surface, and final surface once.\nInspect enough public structure to understand the task.\nCreate the first complete runnable artifact early.\nPrefer native structured tools over ad hoc manipulation.\nConvert requirements into final-surface predicates.\nEdit the smallest live write set.\nPreserve public labels, entity boundaries, missing values, and reference-only targets.\nVerify the final artifact or runtime state through the real consumer.\nReport the changed surface and current proof, separating blockers and non-proof.\n```\n\n## Why agents need it\n\nModern agents can already write, browse, edit, run tools, inspect files, call APIs, and produce artifacts. The recurring failure point is not only raw capability; it is loss of discipline at the boundary between task, evidence, edit scope, and proof.\n\nTypical failures include treating partial artifacts as done, pre-seeding names or payloads a future client must create, expanding broad nouns into unrequested members, flattening missing data into zero, creating records from relation-only targets, accepting weak source matches, replacing a requested literal with a syntactic variant, proving an alternate artifact instead of the final surface, mutating history during cleanup, or reporting an old pass as current proof.\n\nVowline gives agents a portable default for those failure points. It makes the agent bind the target, future client, invocation surface, and final surface; keep facts, sources, exceptions, unknowns, decisions, and proof predicates separate; preserve public labels and declared entity boundaries; prefer native structured tools and real consumers; limit writes to the smallest live surface; and verify the delivered surface before declaring completion.\n\nFor the full details, see the canonical [`SKILL.md`](skills\u002Fvowline\u002FSKILL.md).\n\n## Supported agents\n\nVowline keeps the canonical behavior in `skills\u002Fvowline\u002FSKILL.md` and installs host bridge files for tools that read rules, memories, or instruction files.\n\n| Agent \u002F harness | Project install writes |\n| --- | --- |\n| Codex \u002F AGENTS-aware tools | `.agents\u002Fskills\u002Fvowline\u002FSKILL.md`, `AGENTS.md` |\n| Claude Code | `.claude\u002Fskills\u002Fvowline\u002FSKILL.md`, `CLAUDE.md` |\n| Windsurf | `.windsurf\u002Fskills\u002Fvowline\u002FSKILL.md`, `.windsurf\u002Frules\u002Fvowline.md` |\n| Cursor | `.cursor\u002Fskills\u002Fvowline\u002FSKILL.md`, `.cursor\u002Frules\u002Fvowline.mdc` |\n| Gemini CLI | `.gemini\u002Fskills\u002Fvowline\u002FSKILL.md`, `GEMINI.md` |\n| GitHub Copilot | `.github\u002Fskills\u002Fvowline\u002FSKILL.md`, `.github\u002Fcopilot-instructions.md` |\n| Community `SKILL.md` targets | OpenCode, Amp, Goose, Cline, Roo Code, Aider, OpenClaw, Trae skill folders |\n\nCodex uses `~\u002F.agents\u002Fskills\u002Fvowline\u002FSKILL.md` as the documented user-level skill path. The installer also mirrors the skill into `${CODEX_HOME:-~\u002F.codex}\u002Fskills\u002Fvowline\u002FSKILL.md` for compatibility with Codex environments that use `CODEX_HOME\u002Fskills`. Treat that mirror as compatibility support, not as the primary Codex path.\n\nSee [INSTALL.md](INSTALL.md) and [docs\u002FCOMPATIBILITY.md](docs\u002FCOMPATIBILITY.md) for exact global and project paths. See [UPDATE.md](UPDATE.md) and [UNINSTALL.md](UNINSTALL.md) for update and removal flows.\n\n## Scripted install and update\n\nThe scripts are optional. They use only the Python standard library and are safe to run repeatedly.\n\nClone the repository first:\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline.git\ncd vowline\n```\n\nInstall into the current user's agent locations:\n\n```bash\npython3 install.py global --harnesses core\npython3 install.py verify-global --harnesses core\n```\n\nUpdate an existing install by pulling the latest repository and running the same install command again:\n\n```bash\ngit pull\npython3 install.py global --harnesses core\npython3 install.py verify-global --harnesses core\n```\n\nInstall into one project:\n\n```bash\npython3 install.py project \u002Fpath\u002Fto\u002Fproject --harnesses core\npython3 install.py verify \u002Fpath\u002Fto\u002Fproject --harnesses core\n```\n\nInstall only selected harnesses:\n\n```bash\npython3 install.py project \u002Fpath\u002Fto\u002Fproject --harnesses codex,claude,windsurf\npython3 install.py global --harnesses codex\n```\n\nHarness groups:\n\n| Group | Includes |\n| --- | --- |\n| `core` | Codex, Claude Code, Windsurf, Cursor, Gemini CLI, GitHub Copilot |\n| `community` | OpenCode, Amp, Goose, Cline, Roo Code, Aider, OpenClaw, Trae |\n| `all` | `core` plus `community` |\n\nThe script default is `all`. Pass `--harnesses core` when you want only the mainstream host targets.\n\n## How to use it\n\nAfter installation, direct invocation is usually optional because host bridge files activate or route to Vowline where the host supports that pattern. Direct invocation is still useful when you want to force the skill for one request:\n\n```text\n$vowline repair this CLI and verify it through the public command surface\n\u002Fvowline build this spreadsheet with real formulas and inspect the reopened workbook\n@vowline audit this repository for active credential exposure only\n$vowline produce this source-derived report without inventing unsupported rows\n```\n\nCodex commonly uses `$vowline`, Claude Code commonly uses `\u002Fvowline`, and Windsurf commonly uses `@vowline`. Other hosts may surface skills or rules differently.\n\n## Repository layout\n\n```text\nskills\u002Fvowline\u002FSKILL.md         Canonical Vowline contract\nguidance\u002FVOWLINE_ACTIVATION.md  Canonical activation bridge body\ninstall.py                      Optional installer and verifier\nuninstall.py                    Optional uninstaller and verifier\nUPDATE.md                       Update guide for existing installs\ndocs\u002FCOMPATIBILITY.md           Supported paths and host notes\ndocs\u002Fbenchmarks\u002F                Benchmark summaries and charts\ntests\u002F                          Installer, verifier, and uninstall coverage\n```\n\nWhen Vowline edits an existing instruction file, it uses a marked block and preserves unrelated content. The activation body has one source of truth: [guidance\u002FVOWLINE_ACTIVATION.md](guidance\u002FVOWLINE_ACTIVATION.md).\n\n```text\n\u003C!-- vowline:start -->\n\u003Ccontents of guidance\u002FVOWLINE_ACTIVATION.md>\n\u003C!-- vowline:end -->\n```\n\nRepeated installs replace only that marked block. Generated rule files for Cursor and Windsurf use the same activation body with host-required front matter added by `install.py`. Manual installers can render the exact block or rule text with `python3 install.py render-guidance marked-block`, `python3 install.py render-guidance CURSOR.mdc`, or `python3 install.py render-guidance WINDSURF.md`.\n\nExisting installs update by repeating the relevant install command from the latest repository. See [UPDATE.md](UPDATE.md).\n\n## Development checks\n\nRun the same basic checks used by the repository:\n\n```bash\npython3 -m py_compile install.py uninstall.py\npython3 -m unittest discover -s tests\n```\n\nThe tests cover project installs, global installs, harness selection, idempotent marked-block replacement, legacy cleanup, verification, and uninstall behavior.\n\n## Uninstall\n\nAsk the agent to remove Vowline:\n\n```text\nUninstall Vowline from yourself by following https:\u002F\u002Fgithub.com\u002Fchojondocho\u002Fvowline\u002Fblob\u002Fmain\u002FUNINSTALL.md. Verify removal.\n```\n\nOr use the optional script:\n\n```bash\npython3 uninstall.py global --harnesses core\npython3 uninstall.py verify-global --harnesses core\n```\n\nFor project-local removal:\n\n```bash\npython3 uninstall.py project \u002Fpath\u002Fto\u002Fproject --harnesses core\npython3 uninstall.py verify \u002Fpath\u002Fto\u002Fproject --harnesses core\n```\n\nUninstall removes Vowline-owned skill directories and Vowline marked blocks. It should not remove unrelated user or project instructions.\n\n## Name\n\n`Vowline` means a line of commitment: a small covenant tying the agent to the user's public outcome. It is not legal trademark clearance.\n\n## License\n\nMIT. See [LICENSE](LICENSE).\n","Vowline 是一个为AI代理设计的紧凑操作契约。它通过定义一种通用的操作技能，使得AI能够在执行任务时绑定公共任务、依据证据行动、保持边界并验证交付结果。该项目采用Python编写，支持Codex、Claude Code等多种技能兼容的代理平台，并遵循MIT许可证发布。Vowline特别适用于需要明确交付物、约束条件、证据要求以及完成标准的实际工作中，确保在处理数据解释、工件保真度等方面的一致性和准确性。此外，Vowline不会覆盖更高优先级的系统、工具安全或用户明确指示，从而保证了操作的安全性和合规性。",2,"2026-06-11 04:01:46","CREATED_QUERY"]