The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes an installation option that downloads a shell script from the author's GitHub repository and pipes it directly to the bash shell for execution: curl -fsSL https://raw.githubusercontent.com/machina-sports/machina-cli/main/install.sh | bash. This is a vendor-owned resource used for legitimate installation.
[EXTERNAL_DOWNLOADS]: Provides instructions to download and install the machina-cli package from the Python Package Index using pip install machina-cli.
[COMMAND_EXECUTION]: The skill requires the agent to execute several local shell commands to manage authentication and project configuration, such as machina login, machina project use, and machina auth whoami.
[PROMPT_INJECTION]: The skill functions by ingesting external data from a hosted MCP server (fixtures, market prices, and social sentiment), creating a surface for potential Indirect Prompt Injection.
Ingestion points: Data returned by tools like worldcup-search-markets and worldcup-fan-sentiment-context (as described in SKILL.md).
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands in the tool outputs.
Capability inventory: The skill interacts with the local system via machina-cli subprocess calls.
Sanitization: There is no documentation of sanitization or validation performed on the external data before it is presented to the agent.

world-cup