wzh4464

Evaluate agent-generated code against ground truth diff and handwritten file list. Prefer reading GT inputs directly from base_repo via experiment metadata. Trigger on "/diff-eval-local", "evaluate diff", "eval experiment".

Full dev workflow - worktree setup, commit, push, PR, and post-merge cleanup. Use when starting feature work, committing, opening PRs, or cleaning up after merge.

Use when downloading models from Hugging Face, especially large models (LLM, diffusion, VLM) that fail with timeouts or connection drops. Trigger on "download model", "hf download", "pull model from Hugging Face", or when hf download crashes with httpx.ReadTimeout / RemoteProtocolError.

Evaluate agent-generated code changes against a human-approved PR with deterministic file coverage and function-level coverage analysis. Extracts function/symbol changes from diff hunk headers and compares at granular level. Use when evaluating agent code generation quality, benchmarking AI coding tools, or comparing generated patches against human-approved pull requests. Trigger on "/diff-eval-func", "evaluate diff with function coverage", "function-level diff eval".

从论文中提取图片，优先从arXiv源码包获取真正的论文图

深度分析单篇论文，生成详细笔记和评估，图文并茂

在已整理的论文笔记中搜索相关内容

每日研究工作流启动 - 生成论文推荐 + AI 行业动态笔记

Audit an experiment run (agent+prompt+task) for suspicious output that may have come from the actual GT patch or upstream PR rather than independent implementation. Run whenever a new agent run is declared PASS, or whenever an eval report looks suspiciously high-coverage.

对指定 task×type 组合，并行运行 4 个 agent（Claude/Cursor/Codex/OpenCode）的标准跑法。包含 base_repo 清洁验证、eval 保护、CONSTRAINT_DIRECTIVE 注入、实验目录创建、各 agent 启动命令。

使用 Claude Code agent team 批量评测指定目录中的实验。并行评测，报告名称包含 "claude"。Trigger on "/diff-eval-claude", "claude eval", "用 claude 评测".

使用 Codex CLI (默认 OpenAI 订阅，无需额外配置) 批量评测实验目录。顺序或并行运行，报告名称包含 "codex"。Trigger on "/diff-eval-codex", "codex eval", "用 codex 评测".

Evaluate agent-generated code against ground truth diff and handwritten file list. Prefer reading GT inputs directly from base_repo via experiment metadata. Trigger on "/diff-eval-local", "evaluate diff", "eval experiment".

使用 OpenCode CLI 以指定模型批量评测实验目录。报告名称包含 "opencode-<model>"。Trigger on "/diff-eval-opencode", "opencode eval", "用 opencode 评测".

Launch a multi-agent team to execute benchmark tasks in parallel. Team lead handles all setup (repo copy, hooks, metadata); agents directly code in their sessions without spawning subprocess claude invocations.

Acquires missing code repositories for the selected idea (Phase A) and conducts comprehensive code survey mapping academic concepts to implementations (Phase B). Outputs acquired_code_repos, updated_prepare_res, and model_survey for downstream use by inno-implementation-plan.

Evaluate agent-generated code changes against a human-approved PR. Compares the local git diff (modified + untracked files, relative to the PR's merge base commit) with the ground truth PR diff, filters out auto-generated files, scores using a structured rubric (functional correctness, completeness, behavioral equivalence), and provides both semantic and data-based coverage analysis. Use when evaluating agent code generation quality, benchmarking AI coding tools, or comparing generated patches against human-approved pull requests. Trigger on "/diff-eval", "evaluate diff", "compare diff with PR", "score the changes".