Claude Opus 4.7 (max)

adindamochamad / OmniBridge 公开的代码代理与软件工程案例，来源为 公开代码库，复核于 2026-06-27T08:40:26Z。 Model Atlas 将它标记为 A 类证据，因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务：这是一个围绕软件工程任务执行的真实任务，公开材料可以回溯到具体使用者和具体产物。 原始资料写作：An AI agent for backend/IoT engineers that analyzes legacy serial device traffic, including binary, undocumented, or proprietary protocols, to infer protocol structure and integration logic.

公开产物：公开材料提供公开代码、README 或项目配置，可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作：The repository presents a Tauri/Svelte/Rust application with tests and a public demo video, claiming protocol identification in under a minute for industrial integration workflows.

模型作用：Claude Opus 4.7 (max) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作：The README explicitly labels the project as built for the Claude Opus 4.7 Hackathon 2026 and displays Claude Opus 4.7 as the model layer for reasoning about raw bytes and protocol behavior.

A 类理由：有具体使用者、具体任务、公开原始证据和可访问产物，可公开核验。

风险边界：当前判断基于公开材料；若产物下线、仓库变更或模型参与比例仅来自作者自述，需要在引用前重新复核。 原始资料写作：Evidence is self-reported in a public hackathon repository; claims such as speed and protocol coverage should be treated as project claims rather than independently benchmarked results.

原始记录：OmniBridge uses Claude Opus 4.7 to identify undocumented legacy serial-device protocols

bedriyan / Medkit 公开的代码代理与软件工程案例，来源为 公开代码库，复核于 2026-06-27T08:40:26Z。 Model Atlas 将它标记为 A 类证据，因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务：这是一个围绕软件工程任务执行的真实任务，公开材料可以回溯到具体使用者和具体产物。 原始资料写作：A browser-based ER and polyclinic clinical training simulator where learners speak with AI patients, order tests, diagnose, treat, and receive post-session assessment.

公开产物：公开材料提供公开代码、README 或项目配置，可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作：The public repository describes an OSCE-style simulator that produces AI patient interactions and an attending-style grading report covering communication, history-taking, and clinical reasoning with guideline citations.

模型作用：Claude Opus 4.7 (max) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作：The README states that an attending physician powered by Claude Opus 4.7 watches and grades decisions after each session, and that the project was built using Claude Code with Opus 4.7.

A 类理由：有具体使用者、具体任务、公开原始证据和可访问产物，可公开核验。

风险边界：当前判断基于公开材料；若产物下线、仓库变更或模型参与比例仅来自作者自述，需要在引用前重新复核。 原始资料写作：Evidence is from the project author's public GitHub README; it is a hackathon/open-source artifact, not an independently audited customer story. Medical cases are described as synthetic and not clinical claims.

原始记录：Medkit uses Claude Opus 4.7 as an attending grader for browser-based clinical training

leventilo / Mobius 公开的代码代理与软件工程案例，来源为 公开代码库，复核于 2026-06-27T08:40:26Z。 Model Atlas 将它标记为 A 类证据，因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务：这是一个围绕软件工程任务执行的真实任务，公开材料可以回溯到具体使用者和具体产物。 原始资料写作：A system where a user drops in a physics paper PDF and receives a Ciechanowski-style interactive HTML simulator based on the paper's physics.

公开产物：公开材料提供公开代码、README 或项目配置，可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作：The repository describes a pipeline that parses the paper, infers physics, composes a typed SimSpec, generates Python solver primitives, runs them in an Anthropic Managed Agents sandbox, and renders an interactive scene…

模型作用：Claude Opus 4.7 (max) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作：The README states that an Opus 4.7 orchestrator drives a DAG of nine specialist skills and lists the model as claude-opus-4-7 in the stack.

A 类理由：有具体使用者、具体任务、公开原始证据和可访问产物，可公开核验。

风险边界：当前判断基于公开材料；若产物下线、仓库变更或模型参与比例仅来自作者自述，需要在引用前重新复核。 原始资料写作：Evidence is from a public GitHub README for a hackathon/MVP artifact; live demo links are marked pending, so the repository itself is used as the accessible artifact.

原始记录：Mobius uses a Claude Opus 4.7 orchestrator to turn physics papers into interactive simulators

Saadzwak / Archoff 公开的3D 与 Web 交互案例，来源为 公开代码库，复核于 2026-06-27T08:40:26Z。 Model Atlas 将它标记为 A 类证据，因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务：这是一个围绕浏览器 3D 世界构建的真实任务，公开材料可以回溯到具体使用者和具体产物。 原始资料写作：A co-architect product for office interiors that turns a client brief and floor plan into a sourced functional program, 3D test-fit variants, mood board, client argumentaire, slide deck, report, and DXF export.

公开产物：公开材料提供公开代码、README 或项目配置，可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作：The repository describes outputs including costed programs, three 3D SketchUp variants, A3 mood-board PDFs, an 18-slide pitch deck, an A4 report, and a dimensioned A1 DXF with named layers.

模型作用：Claude Opus 4.7 (max) 在该案例中承担浏览器 3D 世界构建相关的生成、分析、编排或实现角色。 原始资料写作：The README states that everything is orchestrated by Claude Opus 4.7, with Vision HD reading plans, managed-agent orchestration producing program/variants/argumentaire, and MCP resources backing decisions.

A 类理由：有具体使用者、具体任务、公开原始证据和可访问产物，可公开核验。

风险边界：当前判断基于公开材料；若产物下线、仓库变更或模型参与比例仅来自作者自述，需要在引用前重新复核。 原始资料写作：Evidence is from the public project README and screenshots in the repo; it is a hackathon/open-source implementation rather than a verified commercial deployment.

原始记录：Archoff uses Claude Opus 4.7 to generate office interior programs, 3D test fits, pitch decks, and DXF exports

srichandrak / Maieutic 公开的代码代理与软件工程案例，来源为 公开代码库，复核于 2026-06-27T08:40:26Z。 Model Atlas 将它标记为 A 类证据，因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务：这是一个围绕软件工程任务执行的真实任务，公开材料可以回溯到具体使用者和具体产物。 原始资料写作：A pedagogical IDE for programming education that asks students to specify intended behavior before coding, supports reasoning-oriented coding conversations, compares submitted code against the student's specification, a…

公开产物：公开材料提供公开代码、README 或项目配置，可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作：The public repository describes a Next.js classroom app with student and teacher flows, live dashboard summaries, specification clarification, code/spec comparison, and teacher-facing class pattern detection.

模型作用：Claude Opus 4.7 (max) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作：The README lists @anthropic-ai/sdk with model claude-opus-4-7 and states that Opus is called at seven moments across student and teacher experiences, including spec clarification, coding guidance, code/spec comparison, …

A 类理由：有具体使用者、具体任务、公开原始证据和可访问产物，可公开核验。

风险边界：当前判断基于公开材料；若产物下线、仓库变更或模型参与比例仅来自作者自述，需要在引用前重新复核。 原始资料写作：Evidence is from the author's public GitHub README; it is an MVP-style educational tool and not a third-party deployment report.

原始记录：Maieutic uses Claude Opus 4.7 to coach programming students and summarize learning signals for teachers