Move agents/skills -> .agents/skills (post-release; resolve the bench-branch divergence with it)
维护者通常 1 天内回复
还没有人认领这个 Issue。
评估
调研方向
Read AGENTS.md, CLAUDE.md, the self-references in review/SKILL.md and plan/SKILL.md, and compare the skill copies on master and bench/gfql-pokec-polars-lanes. The issue says the bench branch’s +52-line review-protocol edit has already landed via #1924; reconcile the copies, update the documented paths, and verify a fresh clone exposes skills to Codex and Claude Code without manual steps.
由索引模型根据 Issue 内容生成。
描述
Owner decision: .agents/skills/ is the universal convention and pygraphistry should adopt it. Post-release — PRs stay on the current agents/ layout until the GFQL stack lands.
Current state (measured)
- master tracks
agents/skills/— 3 SKILL.md files. bench/gfql-pokec-polars-lanestracks.agents/skills/and has noagents/. The rename happened in02fc3d1b1, an unrelated benchmark commit — an accidental divergence, not a deliberate migration. It will conflict or double-apply if that branch merges, so fold it into this work rather than letting a bench commit define the repo layout.CLAUDE.mdcurrently documents the opposite of the target ("Canonical repo skills are inagents/skills/", plus aln -s agents .agentsstep).
Blast radius: small — 8 references in 4 files
AGENTS.md, CLAUDE.md, and self-references inside the skills (review/SKILL.md ×3, plan/SKILL.md ×1), e.g. agents/skills/review/SKILL.md#vectorization--engine-compatibility. Those in-document anchors are the easy miss because they still read correctly after a rename. No CI or bin/ script references the path, so this cannot break a lane — only documentation links.
Plan (single PR)
git mv agents .agentsas one rename-only commit so--find-renameskeeps history legible.- Fix the 8 references; then
git grep agents/skillsmust return zero hits that are not.agents/skills. - Rewrite CLAUDE.md's "Skill path conventions" block:
.agents/skills/canonical, keep the.claude/skillssymlink step for Claude Code, drop the now-circularln -s agents .agents. - Reconcile the bench branch: diff the two copies and fold in anything it added (it currently carries a +52-line review-protocol edit that was applied there by mistake and has since been landed on master properly via #1924).
- Verify a fresh clone gives both Codex and Claude Code a working skills path with no undocumented manual step.
Why not now
A tree-wide rename conflicts with every open branch touching those files. ~15 stacked PRs are open and #1924 is pending against agents/skills/review/SKILL.md; renaming first would force a rebase for zero benefit.
- 主要语言
- Python
- 星标
- 2.6k
- 派生
- 229
- 平均合并
- 1 天 5 小时
- 30 天内合并 PR
- 99
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
graphistry/pygraphistry 的其他 Issue
-
难度 2/5 1-3 小时 新手友好度 62/100
graphistry/pygraphistry#2171 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 68/100
graphistry/pygraphistry#1965 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 65/100
graphistry/pygraphistry#1208 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 72/100
graphistry/pygraphistry#966 · 1 条评论 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 68/100
graphistry/pygraphistry#824 ·
维护者通常 1 天内回复
查看 graphistry/pygraphistry 的全部 Issue
相似的 Issue
-
ci-rerun: collect.py fails a whole pr-babysit cycle on one transient GitHub 5xx and never retries未关闭follow-up
难度 2/5 1-3 小时 新手友好度 62/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 70/100
-
Built-in QUBO fallback applies bitwise xor to floating binary states可能已有人在做 @iloveeyjafjalla 今天认领。 未关闭
难度 2/5 1-3 小时 新手友好度 72/100
qboson/kaiwu-pytorch-plugin#1702 ·
维护者通常 3 天内回复
-
bug v2
难度 2/5 1-3 小时 新手友好度 76/100
modelcontextprotocol/python-sdk#3811 ·
维护者通常 1 天内回复
-
cost:cheap practicality:seen-in-the-wild severity:medium
难度 2/5 1-3 小时 新手友好度 72/100
fairagro/m4.2_advanced_middleware_api#620 ·
维护者通常 1 天内回复