The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill downloads and installs a versioned runtime environment from the author's GitHub repository. Additionally, the skillify module synthesizes and executes new TypeScript scripts based on recorded browser interactions to automate data extraction.
[COMMAND_EXECUTION]: The system automatically detects the project's programming language and executes shell commands to install dependencies (e.g., npm install, gem install, pip install) and run various test frameworks.
[DATA_EXFILTRATION]: The pair-agent module utilizes the ngrok service to establish secure tunnels, exposing a local HTTP server to the internet to allow remote agents to share the browser session. While auth-gated and using temporary keys, this involves intentional external network exposure of a local service.
[PROMPT_INJECTION]: The skill processes large volumes of untrusted data from the public web, browser consoles, and network logs. It implements a defensive posture using strict boundary markers and instructions to ignore any commands or code found within those boundaries.
[EXTERNAL_DOWNLOADS]: A bootstrap utility fetches manifests and component archives from the author's official GitHub releases. These downloads are subject to integrity checks including SHA-256 validation and Cosign release attestations.

qa