microsoft/hve-core
Publicmirrored fromhttps://github.com/microsoft/hve-coreAvailable
docs/getting-started/methods/cli-plugins.md
182lines · modecode
| 1 | --- |
| 2 | title: Copilot CLI Plugins |
| 3 | description: Install HVE Core agents, prompts, and skills as Copilot CLI plugins |
| 4 | sidebar_position: 2 |
| 5 | author: Microsoft |
| 6 | ms.date: 2026-03-23 |
| 7 | ms.topic: how-to |
| 8 | --- |
| 9 | |
| 10 | Install HVE Core collections as Copilot CLI plugins for terminal-based |
| 11 | AI-assisted development workflows. |
| 12 | |
| 13 | ## Prerequisites |
| 14 | |
| 15 | * GitHub Copilot CLI installed and authenticated |
| 16 | * Git symlink support enabled (Windows: Developer Mode + |
| 17 | `git config --global core.symlinks true`) |
| 18 | |
| 19 | ## Register hve-core as a Plugin Marketplace |
| 20 | |
| 21 | ```bash |
| 22 | copilot plugin marketplace add microsoft/hve-core |
| 23 | ``` |
| 24 | |
| 25 | ## Browse Available Plugins |
| 26 | |
| 27 | Type `/plugin` in a Copilot CLI chat session to browse available plugins. |
| 28 | |
| 29 | ## Install a Plugin |
| 30 | |
| 31 | Choose **one** of the following plugins to install. Each command installs a |
| 32 | different collection from the hve-core marketplace. |
| 33 | |
| 34 | For the core Research, Plan, Implement, Review lifecycle: |
| 35 | |
| 36 | ```bash |
| 37 | copilot plugin install hve-core@hve-core |
| 38 | ``` |
| 39 | |
| 40 | For the full bundle (includes everything in `hve-core` plus all additional |
| 41 | collections): |
| 42 | |
| 43 | ```bash |
| 44 | copilot plugin install hve-core-all@hve-core |
| 45 | ``` |
| 46 | |
| 47 | > [!TIP] |
| 48 | > `hve-core-all` is a superset of `hve-core`. Install one or the other, not |
| 49 | > both. If you are unsure which to pick, start with `hve-core-all` for the |
| 50 | > complete experience. |
| 51 | |
| 52 | ## Available Plugins |
| 53 | |
| 54 | | Plugin | Description | |
| 55 | |------------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------| |
| 56 | | hve-core | Research, Plan, Implement, Review lifecycle | |
| 57 | | github | GitHub issue management | |
| 58 | | ado | Azure DevOps integration | |
| 59 | | coding-standards | Language-specific coding guidelines | |
| 60 | | project-planning | PRDs, BRDs, ADRs, architecture diagrams | |
| 61 | | data-science | Data specs, notebooks, dashboards | |
| 62 | | design-thinking | Design thinking coaching and methodology | |
| 63 | | security | Security and incident response | |
| 64 | | installer | Installer skill for guided workspace setup and MCP auto-configuration ([Extension](https://marketplace.visualstudio.com/items?itemName=ise-hve-essentials.hve-installer)) | |
| 65 | | experimental | Experimental and preview artifacts | |
| 66 | | hve-core-all | Full HVE Core bundle | |
| 67 | |
| 68 | ## Plugin Contents |
| 69 | |
| 70 | Each plugin includes: |
| 71 | |
| 72 | | Component | CLI Discovery | Description | |
| 73 | |--------------|---------------|----------------------------------------------------| |
| 74 | | Agents | Yes | Custom chat agents for specialized workflows | |
| 75 | | Commands | Yes | Task prompts accessible via the CLI | |
| 76 | | Skills | Yes | Self-contained skill packages (hve-core-all only) | |
| 77 | | Instructions | No | Included for `#file:` references, not auto-applied | |
| 78 | |
| 79 | Artifacts are symlinked from the plugin directory to the source repository, |
| 80 | enabling zero-copy installation. |
| 81 | |
| 82 | ## Limitations |
| 83 | |
| 84 | ### Instructions are not auto-applied from plugins |
| 85 | |
| 86 | The Copilot CLI [plugin spec](https://docs.github.com/en/copilot/reference/copilot-cli-reference/cli-plugin-reference) |
| 87 | recognizes `agents`, `skills`, `commands`, `hooks`, `mcpServers`, and |
| 88 | `lspServers` as component types. There is no `instructions` component type. |
| 89 | |
| 90 | The CLI loads path-specific instructions exclusively from |
| 91 | `.github/instructions/**/*.instructions.md` in the |
| 92 | [project repo](https://docs.github.com/en/copilot/reference/custom-instructions-support#copilot-cli). |
| 93 | Instruction files in plugin directories are **not** auto-applied via `applyTo` |
| 94 | pattern matching. |
| 95 | |
| 96 | Instruction files are still included in plugin output because agents and |
| 97 | prompts reference them via `#file:` directives. Those cross-file references |
| 98 | resolve correctly within the plugin directory tree. The difference is between |
| 99 | explicit inclusion (an agent pulls in instruction content at execution time) |
| 100 | and automatic application (the CLI matches `applyTo` patterns against the |
| 101 | files you are editing). |
| 102 | |
| 103 | For full path-specific instruction behavior, copy instruction files into your |
| 104 | project's `.github/instructions/` directory. |
| 105 | |
| 106 | ### Other limitations |
| 107 | |
| 108 | * Skills require skill-compatible agent environments |
| 109 | |
| 110 | ## Using Agents After Installation |
| 111 | |
| 112 | After installing a plugin, agents and named commands are available in your CLI session. |
| 113 | |
| 114 | ### Named Commands vs Agent Mode |
| 115 | |
| 116 | CLI plugins provide two distinct interaction patterns: |
| 117 | |
| 118 | | Mode | Command | Behavior | |
| 119 | |---------------|--------------------------|--------------------------------------------------------------| |
| 120 | | Named Command | `/git-commit` | Executes a predefined workflow, then returns to default mode | |
| 121 | | Agent Mode | `/agent Task Researcher` | Switches to the agent for open-ended conversation | |
| 122 | |
| 123 | Named commands (prompts) run a specific workflow and produce structured output. Agent mode enables freeform conversation with a specialized agent until you exit. |
| 124 | |
| 125 | > [!IMPORTANT] |
| 126 | > The CLI does not support prompts that switch to a custom agent directly. |
| 127 | > Prompts like `/task-research` are designed to run within a specific agent |
| 128 | > context. To use them, first switch to the agent, then run the prompt: |
| 129 | > |
| 130 | > ```text |
| 131 | > /agent Task Researcher |
| 132 | > /task-research topic="API authentication patterns" |
| 133 | > ``` |
| 134 | > |
| 135 | > Prompts that do not require an agent context (e.g., `/git-commit`, |
| 136 | > `/git-merge`) work directly from the default mode. |
| 137 | |
| 138 | ### Example: Research Workflow |
| 139 | |
| 140 | Switch to the agent first, then run the prompt: |
| 141 | |
| 142 | ```text |
| 143 | > /agent Task Researcher |
| 144 | Switched to Task Researcher |
| 145 | > /task-research topic="API authentication patterns" |
| 146 | [Agent executes research workflow, creates research document] |
| 147 | ``` |
| 148 | |
| 149 | Continue with follow-up questions in the same agent context: |
| 150 | |
| 151 | ```text |
| 152 | > What are common API authentication patterns for REST APIs? |
| 153 | [Research conversation continues] |
| 154 | > How do OAuth2 and API keys compare for microservices? |
| 155 | [Follow-up within same agent context] |
| 156 | > /exit |
| 157 | ``` |
| 158 | |
| 159 | ### Available Agents |
| 160 | |
| 161 | After installing the hve-core plugin, these agents are available via `/agent <name>`: |
| 162 | |
| 163 | * Task Researcher - deep research and technical investigation |
| 164 | * Task Planner - implementation planning with phased execution |
| 165 | * Task Implementor - code changes following plans |
| 166 | * Memory - persistent context across sessions |
| 167 | * PR Review - pull request analysis and feedback |
| 168 | |
| 169 | For the complete list, run `/help` in a CLI session to see all available commands and agents. |
| 170 | |
| 171 | ### When to Use Each Mode |
| 172 | |
| 173 | * Use **named commands** (`/git-commit-message`, `/git-merge`) directly from default mode for workflows that do not require a custom agent. |
| 174 | * Use **agent mode** (`/agent <name>`) first, then run agent-specific prompts (`/task-research`, `/task-plan`) for structured workflows that need agent context. |
| 175 | * Stay in **agent mode** for exploratory conversations, follow-up questions, or tasks that don't fit a predefined prompt. |
| 176 | |
| 177 | --- |
| 178 | |
| 179 | <!-- markdownlint-disable MD036 --> |
| 180 | *🤖 Crafted with precision by ✨Copilot following brilliant human instruction, |
| 181 | then carefully refined by our team of discerning human reviewers.* |
| 182 | <!-- markdownlint-enable MD036 --> |
| 183 | |