The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from external websites and search engines (via Tavily and agent-browser) to generate reports. This creates an indirect prompt injection surface where adversarial content on a competitor's site could attempt to influence the agent's subsequent actions.
Ingestion points: External websites accessed via infsh/agent-browser and search results from tavily/search-assistant.
Boundary markers: The skill does not define explicit delimiters or warnings for the agent to ignore instructions embedded in the analyzed content.
Capability inventory: The agent can execute various shell commands and run sub-applications via the infsh tool.
Sanitization: No sanitization logic is present to filter malicious instructions from retrieved data.
[COMMAND_EXECUTION]: The skill relies on the infsh CLI to execute tasks, including web browsing, searching, and code execution. This behavior is documented and aligns with the skill's purpose but involves running external processes based on user queries.
[REMOTE_CODE_EXECUTION]: The skill includes a template for generating positioning maps using an infsh/python-executor. Although the provided code is a static visualization script using matplotlib, it utilizes runtime code execution capabilities.
[EXTERNAL_DOWNLOADS]: The documentation points to a remote GitHub repository (raw.githubusercontent.com/inference-sh/skills) for CLI installation instructions and mentions adding remote skills via npx.

competitor-teardown