Skip to main content
在 Manus 中运行任何 Skill
一键导入
GitHub 仓库

jokull-skills

jokull-skills 收录了来自 jokull 的 5 个 skills,并提供仓库级职业覆盖和站内 skill 详情页。

已收集 skills
5
Stars
0
更新
2026-07-20
Forks
0
职业覆盖
3 个职业分类 · 已分类 100%
仓库浏览

这个仓库中的 skills

agent-vocabulary-coach
软件开发工程师

Coach developers to incrementally expand their vocabulary into the next useful wave of precise, pragmatic engineering language, using their agent chat logs as evidence. Use when someone wants to communicate more accurately with coding agents, improve planning language, identify their personal engineering lexicon, adopt useful industry concepts, or find crisp terms for recurring ideas they currently describe indirectly.

2026-07-20
sandbox-builder
软件开发工程师

Designs and builds an agent sandbox for an existing monorepo: a disposable, per-worktree runtime where coding agents can open the app in a real browser, seed scenarios, inspect logs and side effects, and produce evidence without fighting other agents over ports or a shared dev database. Use when asked to build an agent sandbox, give agents a runtime or browser access, set up per-worktree dev environments, isolate parallel agents, or replace 'I ran it locally' with reproducible proof.

2026-07-17
gh-pr-image
软件开发工程师

Embeds local screenshots, GIFs, visual diffs, or proof images into a GitHub PR body or comment without committing image artifacts to the repo, by hosting them as release assets on a per-PR prerelease. Use when a PR needs visual proof from local files, when `gh pr edit` or `gh issue comment` won't upload an image, or when a temporary image host would expire.

2026-07-17
mole
网络与计算机系统管理员

Wraps the `mo` (mole) Mac cleanup CLI with judgment, memory, and follow-through: runs dry-run previews, curates them against remembered safe/risky verdicts for this machine, recalls past sessions and past user decisions, and brainstorms prune targets mole itself doesn't scan. Use when the user asks to free up disk space, clean the Mac, run mole or mo, or prune old project build artifacts.

2026-07-17
roundtable
项目管理专家

Runs a multi-perspective adversarial critique and synthesis pass over a plan, spec, or architecture proposal. Convenes a council of specialist seats — each an independent subagent that cannot see the others — then synthesizes their critiques into a patched artifact. Use when the user asks for a roundtable, council, swarm, adversarial review, downstream impact check, red team, or specialist feedback before implementation; when a design needs stress-testing after requirements are settled; or when the user wants genuine criticism of a plan or piece of writing instead of agreement.

2026-07-17