S
swarm-routing/skills
The swarm-routing org's agent skills: one Claude Code plugin (swarm-routing) and plain Agent Skills for any…
★ 1·更新 2026-11-17
在 GitHub 查看 ↗链接已于 2026-11-17 核查 · HTTP 200 · 打开的是厂商自己的官网
装完之后怎么算装好了
- 装完该看到什么 — 命令正常结束时退出码为 0,终端里没有 "command not found" 或 "permission denied"。
- 一行自检 — 按该仓库 README 里给出的最小用法跑一次;能按 README 的描述输出就是通了。
- 答不出来怎么办 — 这个项目没有写排错说明,去仓库的 Issues 搜 install。
安装命令取自该仓库;以上判据只描述可观察到的结果,不替代项目自己的文档。
- ★ 数取值
- 1 ★ · swarm-routing/skills · 取自 GitHub 搜索 API · (2026-11-19 20:00:48 核对)
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型:开源项目
- 收录与活跃
- 收录于 2026-11-17 · 最近见到 2026-11-20 · 共见到 4 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 1 ★ · 1 个来源 · repo
- 安装
- 安装命令见上方「安装与配置」。
常见问题(答案都来自我们的记录)
它适合谁 / 用来做什么?
我们把它归在这些标签下: Agent 技能。标签来自条目自身,不是我们强加的分类。
谁维护它、还在更新吗?
仓库
swarm-routing/skills;我们收录于 2026-11-17;最近见到 2026-11-20。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。README 原文(英文,未翻译) · 在 GitHub 读完整 README →
# skills
以下摘录自该仓库的 README(片段)
| Group | Badges |
| Build | |
| Release | |
| Trust | |
| Community & docs | |
The swarm-routing org's agent skills, published two ways from one tree:
as the Claude Code plugin swarm-routing, and as plain
Agent Skills that
skills.sh installs into any agent (Claude Code, Codex,
Cursor, Copilot, …). Each skill is a procedure over a script in
swarm-routing/.github tools/,
so the script does the work and the skill says when and in what order.
| Skill | What it does | Script |
| org-repo | set up or update a repo from the org templates: plan, apply, resolve conflicts, check, open the PR | tools/org-sync/org_sync.py |
| fidelity-sheet | write or check a protocol's paper-fidelity sheet (docs/fidelity.md): every mechanism and parameter sourced to the paper, or marked repo choice / unknown; the anchor candidate | tools/fidelity/fidelity.py |
| protocol-review | review a protocol or tool diff against the org golden rules before a PR: the mechanical checks by script, then wire formats, determinism, paired comparisons and bounded structures by judgement | tools/protocol-review/check_invariants.py |
| docs-review | bring a repository's README, AI files and docs/ to the org standard (AntHocNet's): badges, an explanation, the site page and the ns-3 image, a layout that matches the tree, a docs map, no broken links, no unbacked claims | tools/docs-review/check_docs.py |
| pr-shepherd | take open PRs from review to merge: one line per PR and per unresolved finding, each finding verified then fixed (test first) or declined with a reason, every thread replied to and resolved, merged only when CI is green, no thread is open and the user asked, in dependency order | tools/pr-status/pr_status.py |
| lit-search | find new papers (who cites a paper, its references, a narrow search, related work as a last resort) and confirm a reference field by field against Crossref, Semantic Scholar and OpenAlex, or a thesis without a DOI; free sources only, no key needed | tools/lit-search/lit_search.py |
| references | review and set up a repository's docs/references.bib: what it cites, each work confirmed from Crossref or Zenodo, DOI links first, the fidelity Sources table and CITATION.cff references regenerated and checked (swarm-routing/.github#307) | tools/fidelity/fidelity.py, tools/lit-search/lit_search.py, tools/zenodo/zenodo.py |
| token-audit | measure and cut what a session costs in tokens: what it loads before its first tool call (agent files by section, org-managed and repeated text, a budget as a gate; estimates), and where a finished session's tokens went (the first request's context, calls and result bytes per tool, candidates for a script; measured) | this repository's tools/token-audit/token_audit.py |
| bestpractices | the OpenSSF Best Practices badge (bestpractices.dev): register a repository, propose its passing answers from evidence on main (.bestpractices.json or proposal links the maintainer reviews and saves; nothing logs in), keep them current with the weekly drift check, and answer one pasted criterion with a paste-ready justification; owner-only criteria stay the maintainer's | swarm-routing/.github cmd/bestpractices (Go); this repository's tools/tracker-responses/responses.py (tracker counts) |
| Build | |
| Release | |
| Trust | |
| Community & docs | |
The swarm-routing org's agent skills, published two ways from one tree:
as the Claude Code plugin swarm-routing, and as plain
Agent Skills that
skills.sh installs into any agent (Claude Code, Codex,
Cursor, Copilot, …). Each skill is a procedure over a script in
swarm-routing/.github tools/,
so the script does the work and the skill says when and in what order.
| Skill | What it does | Script |
| org-repo | set up or update a repo from the org templates: plan, apply, resolve conflicts, check, open the PR | tools/org-sync/org_sync.py |
| fidelity-sheet | write or check a protocol's paper-fidelity sheet (docs/fidelity.md): every mechanism and parameter sourced to the paper, or marked repo choice / unknown; the anchor candidate | tools/fidelity/fidelity.py |
| protocol-review | review a protocol or tool diff against the org golden rules before a PR: the mechanical checks by script, then wire formats, determinism, paired comparisons and bounded structures by judgement | tools/protocol-review/check_invariants.py |
| docs-review | bring a repository's README, AI files and docs/ to the org standard (AntHocNet's): badges, an explanation, the site page and the ns-3 image, a layout that matches the tree, a docs map, no broken links, no unbacked claims | tools/docs-review/check_docs.py |
| pr-shepherd | take open PRs from review to merge: one line per PR and per unresolved finding, each finding verified then fixed (test first) or declined with a reason, every thread replied to and resolved, merged only when CI is green, no thread is open and the user asked, in dependency order | tools/pr-status/pr_status.py |
| lit-search | find new papers (who cites a paper, its references, a narrow search, related work as a last resort) and confirm a reference field by field against Crossref, Semantic Scholar and OpenAlex, or a thesis without a DOI; free sources only, no key needed | tools/lit-search/lit_search.py |
| references | review and set up a repository's docs/references.bib: what it cites, each work confirmed from Crossref or Zenodo, DOI links first, the fidelity Sources table and CITATION.cff references regenerated and checked (swarm-routing/.github#307) | tools/fidelity/fidelity.py, tools/lit-search/lit_search.py, tools/zenodo/zenodo.py |
| token-audit | measure and cut what a session costs in tokens: what it loads before its first tool call (agent files by section, org-managed and repeated text, a budget as a gate; estimates), and where a finished session's tokens went (the first request's context, calls and result bytes per tool, candidates for a script; measured) | this repository's tools/token-audit/token_audit.py |
| bestpractices | the OpenSSF Best Practices badge (bestpractices.dev): register a repository, propose its passing answers from evidence on main (.bestpractices.json or proposal links the maintainer reviews and saves; nothing logs in), keep them current with the weekly drift check, and answer one pasted criterion with a paste-ready justification; owner-only criteria stay the maintainer's | swarm-routing/.github cmd/bestpractices (Go); this repository's tools/tracker-responses/responses.py (tracker counts) |
收录徽章(可选)
想把收录标记放进 README 就复制这行,可选,不影响收录。
[](https://anclum.com/zh/s/swarm-routing-skills/)
- 来源
- github · https://github.com/swarm-routing/skills
- 按领域浏览
- Agent 技能
- Traction
- 1 ★
- 收录于
- 今天
更多条目
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。