Skills

Dockerfile

@@ -41,6 +41,16 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     libgbm1 \
     libxshmfence1 \
     libasound2 \
+    unzip \
+    p7zip-full \
+    bc \
+    ripgrep \
+    fd-find \
+    sqlite3 \
+    libsqlite3-dev \
+    wkhtmltopdf \
+    poppler-utils \
+    default-jre \
  && apt-get clean && rm -rf /var/lib/apt/lists/*
 
 
@@ -64,6 +74,12 @@ COPY ./server.py /app/server.py
 # Create application/jupyter directories
 RUN mkdir -p /app/uploads /app/jupyter_runtime
 
+# Copy skills directory structure into the container
+# Public skills are baked into the image
+# User skills directory is created as mount point for user-added skills
+COPY ./skills/public /app/uploads/skills/public
+RUN mkdir -p /app/uploads/skills/user
+
 # # Generate SSH host keys
 # RUN ssh-keygen -A
 
@@ -81,20 +97,26 @@ EXPOSE 8222
 # Start the FastAPI application
 # CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8002", "--workers", "1", "--no-access-log"]
 
+RUN apt-get --fix-broken install
+# Ensure Node.js, npm (and npx) are set up
+RUN curl -fsSL https://deb.nodesource.com/setup_22.x | bash -
+RUN apt-get install -y nodejs
+
+
+
+ENV PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"
+RUN npm install playwright@1.53.0 -g
+RUN npx playwright@1.53.0 install
 
 # Copy the entrypoint script into the image
 COPY entrypoint.sh /entrypoint.sh
 
 # Make the entrypoint script executable
 RUN chmod +x /entrypoint.sh
 
-# Ensure Node.js, npm (and npx) are set up
-RUN curl -fsSL https://deb.nodesource.com/setup_22.x | bash -
-RUN apt-get install -y nodejs
 
-ENV PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"
-RUN npm install playwright@1.53.0 -g
-RUN npx playwright@1.53.0 install
+
+
 
 
 # Use the entrypoint script

README.md

@@ -178,11 +178,77 @@ Code runs in an isolated container with VM-level isolation. Your host system and
 From [@apple/container](https://github.com/apple/container/blob/main/docs/technical-overview.md):
 >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.
 
+## Skills System
+
+CodeRunner includes a built-in skills system that provides pre-packaged tools for common tasks. Skills are organized into two categories:
+
+### Built-in Public Skills
+
+The following skills are included in every CodeRunner installation:
+
+- **pdf-text-replace** - Replace text in fillable PDF forms
+- **image-crop-rotate** - Crop and rotate images
+
+### Using Skills
+
+Skills are accessed through MCP tools:
+
+```python
+# List all available skills
+result = await list_skills()
+
+# Get documentation for a specific skill
+info = await get_skill_info("pdf-text-replace")
+
+# Execute a skill's script
+code = """
+import subprocess
+subprocess.run([
+    'python',
+    '/app/uploads/skills/public/pdf-text-replace/scripts/replace_text_in_pdf.py',
+    '/app/uploads/input.pdf',
+    'OLD TEXT',
+    'NEW TEXT',
+    '/app/uploads/output.pdf'
+])
+"""
+result = await execute_python_code(code)
+```
+
+### Adding Custom Skills
+
+Users can add their own skills to the `~/.coderunner/assets/skills/user/` directory:
+
+1. Create a directory for your skill (e.g., `my-custom-skill/`)
+2. Add a `SKILL.md` file with documentation
+3. Add your scripts in a `scripts/` subdirectory
+4. Skills will be automatically discovered by the `list_skills()` tool
+
+**Skill Structure:**
+```
+~/.coderunner/assets/skills/user/my-custom-skill/
+├── SKILL.md              # Documentation with usage examples
+└── scripts/              # Your Python/bash scripts
+    └── process.py
+```
+
+### Example: Using the PDF Text Replace Skill
+
+```bash
+# Inside the container, execute:
+python /app/uploads/skills/public/pdf-text-replace/scripts/replace_text_in_pdf.py \
+    /app/uploads/tax_form.pdf \
+    "John Doe" \
+    "Jane Smith" \
+    /app/uploads/tax_form_updated.pdf
+```
+
 ## Architecture
 
 CodeRunner consists of:
 - **Sandbox Container:** Isolated execution environment with Jupyter kernel
 - **MCP Server:** Handles communication between AI models and the sandbox
+- **Skills System:** Pre-packaged tools for common tasks (PDF manipulation, image processing, etc.)
 
 ## Examples
 

SKILLS-README.md

@@ -0,0 +1,117 @@
+# Skills powered by coderunner, running locally on your Mac
+
+> [!NOTE]
+> [CodeRunner](https://github.com/instavm/coderunner) executes AI-generated code in a truly isolated sandboxed environment on your Mac using Apple's native containers.
+
+# Pre-requisite
+* `Mac` with a `M-series` chip.
+* Install the latest`coderunner` by running the `./install.sh` script from the main repository.
+```shell
+./install.sh
+```
+
+# How To Use Skills
+* `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.
+
+*For example, for Gemini CLI, you can edit* `~/.gemini/settings.json`
+  ```json
+  {
+  "theme": "Default",
+  "selectedAuthType": "oauth-personal",
+  "mcpServers": {
+    "coderunner": {
+      "httpUrl": "http://coderunner.local:8222/mcp"
+    }
+  }
+}
+```
+
+ And for system instructions, replace the `~/.gemini/GEMINI.md` with the [GEMINI.md](https://github.com/instavm/coderunner/examples/gemini/GEMINI.md)
+
+
+# How To Add New Skills
+
+## Option 1: Import from Claude
+
+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>`
+
+For example, I have added 4 skills in the user folder as:
+```shell
+/Users/manish/.coderunner/assets/skills/
+├── public
+│   ├── image-crop-rotate
+│   │   ├── scripts
+│   │   └── SKILL.md
+│   └── pdf-text-replace
+│       ├── scripts
+│       └── SKILL.md
+└── user
+    ├── docx
+    │   ├── docx-js.md
+    │   ├── LICENSE.txt
+    │   ├── ooxml
+    │   ├── ooxml.md
+    │   ├── scripts
+    │   └── SKILL.md
+    ├── pptx
+    │   ├── html2pptx.md
+    │   ├── LICENSE.txt
+    │   ├── ooxml
+    │   ├── ooxml.md
+    │   ├── scripts
+    │   └── SKILL.md
+    ├── slack-gif-creator
+    │   ├── core
+    │   ├── LICENSE.txt
+    │   ├── requirements.txt
+    │   ├── SKILL.md
+    │   └── templates
+    └── xlsx
+        ├── LICENSE.txt
+        ├── recalc.py
+        └── SKILL.md
+```
+
+
+## Option 2: Write Your Own Skills
+
+* 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)
+* You can also ask claude to generate one like `Can you write a skill which creates ascii art of words given one.`
+  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`
+
+Test drive with Gemini CLI
+
+```
+> /mcp
+
+Configured MCP servers:
+
+🟢 coderunner - Ready (5 tools)
+  Tools:
+  - execute_python_code
+  - get_skill_file
+  - get_skill_info
+  - list_skills
+  - navigate_and_get_all_visible_text
+
+> can you generate ascii art for "CODERUNNER"
+
+✦ I will generate the ASCII art you desire. First, I must survey my available skills.
+✓  list_skills (coderunner MCP Server)
+
+✦ I have located a relevant skill: ascii-art. I will now retrieve its instructions.
+✓  get_skill_info (coderunner MCP Server) {"skill_name":"ascii-art"}
+✦ Your ASCII art is ready:
+
+   1   ____ ___  ____  _____ ____  _   _ _   _ _   _ _____ ____
+   2  / ___/ _ \|  _ \| ____|  _ \| | | | \ | | \ | | ____|  _ \
+   3 | |  | | | | | | |  _| | |_) | | | |  \| |  \| |  _| | |_) |
+   4 | |__| |_| | |_| | |___|  _ <| |_| | |\\  | |\\  | |___|  _ <
+   5  \____\___/|____/|_____|_| \_\\___/|_| \_|_| \_|_____|_| \_\
+   6
+
+
+
+
+Using: 1 GEMINI.md file | 3 MCP servers (ctrl+t to view)
+```

examples/gemini_cli/GEMINI.md

@@ -0,0 +1,16 @@
+always start answer by calling me lord voldemort.
+
+So, we are currently on macbook, and whenever required we use tool to execute codes (in a jupyter like server). the code is executed in a container (you wouldn't notice but just know this).
+
+The paths on local machine is ~/.coderunner/assets/skills/user is mapped to /app/uploads/skills/user inside container.
+
+~/.coderunner/assets/outputs (in the host machine) is mapped to /app/uploads/outputs inside conatiner. This is where user will puts their files they want to edit like some png, pdf, txt etc. You should also use it to output your artifacts generated.
+
+So that will help whenever we need a file inside a container to work on it via the execute code tool.
+
+There are also "skills" which can do jobs by executing scripts already residing in /app/uploads/skills/<public|user>/<skill-name> . There are tools available to check what skills are avaialble, after checking you can decide wchihc specific skill you wantg to use and then get info about that skill using tool. That will have instructions on how to call execute code with stuff like `!python /path/to/script.py <input> <output>`\
+
+Whenever I ask you to do a task, alwasys check if there are skills available in the list which can do it.
+
+Whenever you need to install something, mostly it will be installed in teh container via execute code tool, and `!pip install pyfiglet` command etc.
+

install.sh

@@ -19,7 +19,7 @@ else
   echo "✅ macOS system detected."
 fi
 
-download_url="https://github.com/apple/container/releases/download/0.3.0/container-0.3.0-installer-signed.pkg"
+download_url="https://github.com/apple/container/releases/download/0.5.0/container-0.5.0-installer-signed.pkg"
 
 # Check if container is installed and display its version
 if command -v container &> /dev/null
@@ -57,7 +57,7 @@ echo "Running: sudo container system dns create local"
 sudo container system dns create local
 
 echo "Running: container system dns default set local"
-container system dns default set local
+container system property set dns.domain local
 
 echo "Starting the Sandbox Container..."
 container system start
@@ -66,12 +66,21 @@ container system start
 echo "Pulling the latest image: instavm/coderunner"
 container image pull instavm/coderunner
 
-echo "→ Ensuring coderunner assets directory…"
+echo "→ Ensuring coderunner assets directories…"
 ASSETS_SRC="$HOME/.coderunner/assets"
-mkdir -p "$ASSETS_SRC"
+mkdir -p "$ASSETS_SRC/skills/user"
+mkdir -p "$ASSETS_SRC/outputs"
 
 # Run the command to start the sandbox container
 echo "Running: container run --name coderunner --detach --rm --cpus 8 --memory 4g instavm/coderunner"
-container run  --volume "$ASSETS_SRC:/app/uploads" --name coderunner --detach --rm --cpus 8 --memory 4g instavm/coderunner
+container run \
+  --volume "$ASSETS_SRC/skills/user:/app/uploads/skills/user" \
+  --volume "$ASSETS_SRC/outputs:/app/uploads/outputs" \
+  --name coderunner \
+  --detach \
+  --rm \
+  --cpus 8 \
+  --memory 4g \
+  instavm/coderunner
 
 echo "✅ Setup complete. MCP server is available at http://coderunner.local:8222/mcp"

requirements.txt

@@ -38,3 +38,48 @@ fastmcp
 openai-agents
 
 playwright==1.53.0
+
+# Data Science Libraries
+pandas
+numpy
+scipy
+scikit-learn
+statsmodels
+
+# Visualization Libraries
+matplotlib
+seaborn
+
+# Image/Video I/O Libraries
+imageio
+imageio-ffmpeg
+
+# File Processing Libraries
+pyarrow
+openpyxl
+xlsxwriter
+xlrd
+pillow
+python-pptx
+python-docx
+pypdf
+pdfplumber
+pypdfium2
+pdf2image
+pdfkit
+tabula-py
+reportlab[pycairo]
+img2pdf
+
+# Math & Computing Libraries
+sympy
+mpmath
+
+# Utilities
+tqdm
+python-dateutil
+pytz
+joblib
+
+# Beautiful Soup for HTML parsing (already used in server.py)
+beautifulsoup4