Skip to content

Commit 814e6cf

Browse files
committed
update skills readme
1 parent 19c78b6 commit 814e6cf

File tree

2 files changed

+555
-22
lines changed

2 files changed

+555
-22
lines changed

SKILLS-README.md

Lines changed: 150 additions & 22 deletions
Original file line numberDiff line numberDiff line change
@@ -1,21 +1,42 @@
1-
# Skills powered by coderunner, running locally on your Mac
1+
# Run Claude Skills Locally on Your Mac (No Cloud Upload Required)
22

3-
> [!NOTE]
4-
> [CodeRunner](https://github.com/instavm/coderunner) executes AI-generated code in a truly isolated sandboxed environment on your Mac using Apple's native containers.
3+
Anthropic recently announced [Skills for Claude](https://www.anthropic.com/news/skills) - reusable folders with instructions, scripts, and resources that make Claude better at specialized tasks. CodeRunner lets you run these skills **entirely on your local machine** in a sandboxed environment.
4+
5+
**What this means:** Claude can now process your files (documents, spreadsheets, presentations, images) using these specialized skills while keeping all data on your Mac. No uploads, complete privacy.
6+
7+
> **About CodeRunner:** [CodeRunner](https://github.com/instavm/coderunner) executes AI-generated code in a truly isolated sandboxed environment on your Mac using Apple's native containers.
8+
9+
## Why Run Skills Locally?
10+
11+
- **Privacy:** Process sensitive documents, financial data, or proprietary files without cloud upload
12+
- **Full Control:** Skills execute in an isolated container with VM-level isolation
13+
- **Compatibility:** Works with Claude Desktop, Gemini CLI, Qwen CLI, or any MCP-compatible tool
14+
- **Extensibility:** Import Anthropic's official skills or create your own custom skills
15+
16+
## Prerequisites
17+
18+
* Mac with Apple Silicon (M1/M2/M3/M4/M5)
19+
* Install the latest `coderunner` by running the `./install.sh` script from the [main repository](https://github.com/instavm/coderunner)
520

6-
# Pre-requisite
7-
* `Mac` with a `M-series` chip.
8-
* Install the latest`coderunner` by running the `./install.sh` script from the main repository.
921
```shell
10-
./install.sh
22+
git clone https://github.com/instavm/coderunner.git
23+
cd coderunner
24+
chmod +x install.sh
25+
sudo ./install.sh
1126
```
1227

13-
# How To Use Skills
14-
* `coderunner` is exposed as an MCP and can be connected to tools like `gemini cli` or `qwen cli` or `claude desktop` or anything that supports MCP. The execution is completely local, done on your Mac.
28+
Installation takes ~2 minutes. The MCP server will be available at `http://coderunner.local:8222/mcp`
1529

16-
*For example, for Gemini CLI, you can edit* `~/.gemini/settings.json`
17-
```json
18-
{
30+
## Setup: Connect Your AI Tool
31+
32+
CodeRunner is exposed as an MCP server and works with any MCP-compatible tool. All execution happens locally on your Mac.
33+
34+
### Example: Gemini CLI Configuration
35+
36+
Edit `~/.gemini/settings.json`:
37+
38+
```json
39+
{
1940
"theme": "Default",
2041
"selectedAuthType": "oauth-personal",
2142
"mcpServers": {
@@ -26,16 +47,49 @@
2647
}
2748
```
2849

29-
And for system instructions, replace the `~/.gemini/GEMINI.md` with the [GEMINI.md](https://github.com/instavm/coderunner/examples/gemini/GEMINI.md)
50+
For system instructions, replace `~/.gemini/GEMINI.md` with [GEMINI.md](https://github.com/instavm/coderunner/blob/main/examples/gemini/GEMINI.md)
51+
52+
### Other Supported Tools
53+
54+
- **Claude Desktop:** See [configuration guide](https://github.com/instavm/coderunner#option-1-claude-desktop-integration)
55+
- **Qwen CLI:** Configure similar to Gemini CLI
56+
- **Any MCP client:** Point to `http://coderunner.local:8222/mcp`
3057

58+
## Example Use Cases
3159

32-
# How To Add New Skills
60+
Once configured, you can ask your AI to:
3361

34-
## Option 1: Import from Claude
62+
- "Create a professional PowerPoint presentation from this markdown outline"
63+
- "Extract all tables from these 10 PDFs and combine into one Excel spreadsheet"
64+
- "Generate ASCII art logo for my project"
65+
- "Fill out this tax form PDF with data from my CSV file"
66+
- "Batch process these 100 images: crop to 16:9 and rotate 90 degrees"
3567

36-
You can either download and copy the folder from Anthropic skills's [github repo](https://github.com/anthropics/skills/) to `~/.coderunner/assets/skills/user/<new-skill-folder>`
3768

38-
For example, I have added 4 skills in the user folder as:
69+
## Adding New Skills
70+
71+
You can extend CodeRunner with additional skills in two ways:
72+
73+
### Option 1: Import Anthropic's Official Skills
74+
75+
Download skills from [Anthropic's skills repository](https://github.com/anthropics/skills/) and copy to:
76+
77+
```
78+
~/.coderunner/assets/skills/user/<new-skill-folder>
79+
```
80+
81+
**Available Official Skills:**
82+
- Microsoft Word (docx)
83+
- Microsoft PowerPoint (pptx)
84+
- Microsoft Excel (xlsx)
85+
- PDF manipulation
86+
- Image processing
87+
- Slack GIF creator
88+
- And more...
89+
90+
### Skills Directory Structure
91+
92+
Here's an example with 4 imported skills:
3993
```shell
4094
/Users/manish/.coderunner/assets/skills/
4195
├── public
@@ -73,13 +127,49 @@ For example, I have added 4 skills in the user folder as:
73127
```
74128

75129

76-
## Option 2: Write Your Own Skills
130+
### Option 2: Create Your Own Skills
131+
132+
Create a folder matching the structure shown above. The only mandatory file is `SKILL.md`. See [Anthropic's skills documentation](https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview) for details.
133+
134+
**Quick Method:**
135+
Ask Claude to generate a skill for you:
136+
137+
> "Can you write a skill which creates ASCII art of words?"
77138
78-
* You can create a folder in the similar structure as above, where only mandatory file is the `SKILL.md`. [Docs](https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview)
79-
* You can also ask claude to generate one like `Can you write a skill which creates ascii art of words given one.`
80-
After it creates the skill, it will let you download a `ZIP` file which you can place directly (no need to expand) in `~/.coderunner/assets/skills/user`
139+
Claude will create the skill and offer a ZIP download. Place the ZIP file directly in `~/.coderunner/assets/skills/user` (no need to expand).
81140

82-
Test drive with Gemini CLI
141+
**Manual Method:**
142+
Create your own skill folder structure:
143+
144+
```
145+
~/.coderunner/assets/skills/user/my-custom-skill/
146+
├── SKILL.md # Documentation and usage examples
147+
├── scripts/ # Your Python/bash scripts
148+
│ └── process.py
149+
└── requirements.txt # (optional) Python dependencies
150+
```
151+
152+
## How Folder Mapping Works
153+
154+
CodeRunner provides a compatibility layer that lets you run Claude's skills locally without modification:
155+
156+
**Path Translation:**
157+
- Claude's path: `/mnt/user-data` → CodeRunner's path: `/app/uploads`
158+
- Skills designed for Claude work locally without any changes
159+
160+
**Accessing Your Local Files:**
161+
- Place files in `~/.coderunner/assets/outputs` on your Mac
162+
- They become available to skills inside the container via volume mounts
163+
- The mapping is automatic - skills can access your files without cloud upload
164+
165+
**Skill Structure:**
166+
- No changes needed to imported skills
167+
- Original folder hierarchy and file organization remain identical
168+
- Import Claude skills and use them directly
169+
170+
## Live Demo: ASCII Art Skill
171+
172+
Here's a real example using Gemini CLI:
83173

84174
```
85175
> /mcp
@@ -115,3 +205,41 @@ Configured MCP servers:
115205
116206
Using: 1 GEMINI.md file | 3 MCP servers (ctrl+t to view)
117207
```
208+
209+
**What happened:**
210+
1. AI discovered available skills via `list_skills`
211+
2. Found the relevant `ascii-art` skill
212+
3. Retrieved skill instructions with `get_skill_info`
213+
4. Executed the skill locally in the sandbox
214+
5. Returned results - all without uploading any data to the cloud
215+
216+
217+
218+
From [@apple/container](https://github.com/apple/container/blob/main/docs/technical-overview.md):
219+
> Each container has the isolation properties of a full VM, using a minimal set of core utilities and dynamic libraries to reduce resource utilization and attack surface.
220+
221+
## Available MCP Tools
222+
223+
When connected, CodeRunner exposes these tools to your AI:
224+
225+
- `execute_python_code` - Execute code in the sandbox
226+
- `get_skill_file` - Read skill files
227+
- `get_skill_info` - Get skill documentation
228+
- `list_skills` - List all available skills
229+
- `navigate_and_get_all_visible_text` - Web scraping with Playwright
230+
231+
## Learn More
232+
233+
- **Main Repository:** [github.com/instavm/coderunner](https://github.com/instavm/coderunner)
234+
- **Anthropic Skills:** [github.com/anthropics/skills](https://github.com/anthropics/skills)
235+
- **Skills Documentation:** [docs.claude.com/skills](https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview)
236+
- **Blog: Building Offline Workspace:** [instavm.io/blog/building-my-offline-workspace-part-2-skills](https://instavm.io/blog/building-my-offline-workspace-part-2-skills)
237+
- **Report Issues:** [github.com/instavm/coderunner/issues](https://github.com/instavm/coderunner/issues)
238+
239+
## Contributing
240+
241+
We welcome contributions! If you create useful skills or improve the coderunner implementation, please share them with the community.
242+
243+
## License
244+
245+
This project is licensed under the Apache 2.0 License - see the [LICENSE](https://github.com/instavm/coderunner/blob/main/LICENSE) file for details.

0 commit comments

Comments
 (0)