nibzard/skillctl

nibzard 资料

最近记录的来源活动: 2026年3月21日 17:27

SkillsMP 收录数据更新: 2026年10月17日 00:09

已收集 skills: 12

GitHub 星标: 0
GitHub Forks: 0

仓库 README

仓库 README 来源：README.md

查看 README 源文件

展开完整 README

# skillctl [![CI](https://img.shields.io/github/actions/workflow/status/nibzard/skillctl/ci.yml?branch=main&label=ci)](https://github.com/nibzard/skillctl/actions/workflows/ci.yml) [![Latest release](https://img.shields.io/github/v/release/nibzard/skillctl?label=release)](https://github.com/nibzard/skillctl/releases/latest) [![License](https://img.shields.io/badge/license-Apache--2.0-blue.svg)](./LICENSE) `skillctl` is a local-first skill manager for the open `SKILL.md` ecosystem. It gives you one clean workflow for: - keeping canonical workspace skills in one place - installing skills from Git, local directories, or archives - projecting them into the roots different agent runtimes actually read - understanding what is active, what changed, and what is safe to update If you use more than one coding agent, `skillctl` is the missing control plane. ## Why Without `skillctl`, skill setups usually drift into some mix of duplicated files, ad-hoc symlinks, unclear precedence, and no reliable history. `skillctl` fixes that by managing: - canonical local skills in `.agents/skills/` - imported skills pinned in `.agents/skillctl.lock` - overlays in `.agents/overlays/` - generated runtime-visible projections - local history, diagnostics, and optional public-only telemetry Important rule: - edit canonical skills or overlays, not generated runtime copies ## Install Install the latest release: ```bash curl -fsSL https://github.com/nibzard/skillctl/releases/latest/download/skillctl-install.sh | sh ``` Install a specific version or custom destination: ```bash curl -fsSL https://github.com/nibzard/skillctl/releases/latest/download/skillctl-install.sh | \ SKILLCTL_VERSION=v0.1.0 SKILLCTL_INSTALL_DIR="$HOME/.local/bin" sh ``` Published release artifacts currently cover: - Linux `x86_64` - macOS `x86_64` - macOS `arm64` - Windows `x86_64` as a zip download Build from source: ```bash cargo build --release --locked ./target/release/skillctl --help ``` ## Quickstart Initialize a workspace: ```bash skillctl init skillctl sync ``` Install a skill from Git: ```bash skillctl install https://github.com/acme/skills.git --name ai-sdk --scope workspace ``` Or install from a local directory: ```bash skillctl install ../shared-skills --interactive ``` Inspect what won and where it projects: ```bash skillctl explain ai-sdk skillctl path ai-sdk skillctl doctor ``` Customize without forking the whole source: ```bash skillctl override ai-sdk $EDITOR .agents/overlays/ai-sdk/SKILL.md skillctl sync ``` Update or pin later: ```bash skillctl update ai-sdk skillctl pin ai-sdk main skillctl rollback ai-sdk <version-or-commit> ``` ## Default Layout After `skillctl init`: ```text repo/ .agents/ skills/ overlays/ skillctl.yaml skillctl.lock ``` The local state store lives in: ```text ~/.skillctl/state.db ``` Imported source snapshots are cached under: ```text ~/.skillctl/store/imports/ ``` ## Core Commands | Command | What it does | |---|---| | `skillctl init` | Create the workspace manifest, lockfile, and default layout | | `skillctl install <source>` | Detect and install skills from Git, local paths, or archives | | `skillctl sync` | Rebuild generated runtime projections | | `skillctl update [skill]` | Check upstream state and recommend a safe next action | | `skillctl list` | Show managed installs, projections, and current state | | `skillctl explain <skill>` | Show the active winner, shadowed candidates, and visibility | | `skillctl path <skill>` | Show canonical, cached, overlay, planned, and projected paths | | `skillctl doctor` | Diagnose drift, conflicts, shadowing, and trust issues | | `skillctl history [skill]` | Show installs, updates, rollbacks, overlays, and cleanup events | | `skillctl override <skill>` | Create or reuse an overlay for a managed import | | `skillctl fork <skill>` | Copy an imported skill into local canonical ownership | | `skillctl telemetry status` | Inspect current telemetry settings | | `skillctl tui` | Open the read-only terminal dashboard | | `skillctl mcp serve` | Run the MCP server with v1 tool parity | Every command also supports `--json`. ## Supported Runtimes `skillctl` currently plans projections for: - Codex - Claude Code - GitHub Copilot - Gemini CLI - Amp - OpenCode Default workspace targets are: - `codex` - `gemini-cli` - `opencode` Add more in `.agents/skillctl.yaml` when needed. ## Typical Workflow 1. Keep local first-party skills in `.agents/skills/`. 2. Install third-party skills with `skillctl install`. 3. Let `skillctl sync` project the active result into runtime roots. 4. Use `skillctl explain` and `skillctl doctor` when behavior is unclear. 5. Use `override` for small changes and `fork` when you want full ownership. ## Telemetry Telemetry is: - enabled by default after a first-run notice - public-source only - content-free - immediately opt-outable with `skillctl telemetry disable` It never includes private repository identifiers, skill contents, arbitrary diffs, or private local paths. ## Troubleshooting Start here: ```bash skillctl doctor skillctl explain <skill> skillctl path <skill> skillctl history <skill> ``` Common cases: - missing skill in one runtime: check `explain` and `path` - stale generated copy: run `skillctl sync` - same-name conflict or shadowing: inspect `explain` - local edits blocking updates: check `doctor` and `update` Full guide: [docs/troubleshooting.md](./docs/troubleshooting.md) ## Exit Codes | Code | Meaning | |---|---| | `0` | success | | `1` | operational error | | `2` | success with warnings | | `3` | validation or conflict failure | | `4` | trust gate blocked | | `5` | interactive input required | ## Development Local checks: ```bash cargo fmt --check cargo clippy --all-targets --all-features -- -D warnings cargo test --locked RUSTDOCFLAGS="-D warnings" cargo doc --locked --no-deps ``` More detail: - user guide: [docs/user-guide.md](./docs/user-guide.md) - product direction: [spec.md](./spec.md) - troubleshooting: [docs/troubleshooting.md](./docs/troubleshooting.md)

正在加载仓库 README…

这个仓库中的 skills

软件开发工程师软件质量保证分析师与测试员

2 个职业分类 · 已分类 100%

软件开发工程师

Release notes helper for update blocker tests.

原文语言：无法判定

2026-03-19

已展示 12 / 12 个已收集 Skill。

用 AI 助手安装

复制这段 Prompt，发给你正在使用的 AI 助手。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/nibzard/skillctl.