The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's setup instructions direct the user or agent to execute a remote script using curl -fsSL https://cli.inference.sh | sh. This pattern is highly insecure as it involves running unverified code directly from an external network source.
[EXTERNAL_DOWNLOADS]: The skill relies on external binary downloads from dist.inference.sh and interaction with the inference.sh domain to function.
[DYNAMIC_EXECUTION]: The core functionality of the skill is the dynamic execution of arbitrary Python code in a remote environment, which is inherently risky if the input is not strictly controlled.
[COMMAND_EXECUTION]: Uses the infsh CLI tool to execute shell commands that interact with external services, sending user-defined code as a payload.
[INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection due to its web scraping capabilities.
Ingestion points: The skill instructions provide examples for scraping external websites using requests and BeautifulSoup (SKILL.md).
Boundary markers: No boundary markers or delimiters are provided to distinguish between data being processed and instructions for the agent.
Capability inventory: The agent has the capability to execute arbitrary code and make network requests via the infsh tool (SKILL.md).
Sanitization: The skill lacks specified sanitization or filtering mechanisms for data retrieved from external sources before it is processed.

python-executor