The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local wrapper script located at /Volumes/SSD2T/code/ai-script. It also directs the agent to use the system utility ffmpeg for audio format conversion and uv for running the local project.
[COMMAND_EXECUTION]: The wrapper script is described as automatically sourcing the user's shell configuration file (~/.zshrc) to load environment variables and API keys into the agent's execution environment.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by ingesting untrusted data from external sources.
Ingestion points: The scrape command (fetching web pages via Firecrawl), search command (web results via Tavily), and scholar command (academic data via OpenAlex) all introduce external, potentially adversarial content into the agent's context.
Boundary markers: The instructions do not define boundary markers or delimiters to isolate untrusted external content from the agent's core instructions.
Capability inventory: The skill has the capability to execute shell commands, write files to the local disk (via the -o output flag), and perform network operations through various AI providers.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from web scraping or searches before it is processed by the agent or passed to subsequent LLM sub-commands.

bfs-ai-script