The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process data from external research and search tools (Tavily and Exa) to fact-check and provide context for press releases. This creates a surface where malicious instructions embedded in web content could influence the agent's output.
Ingestion points: Research data retrieved via belt app run commands (e.g., Tavily/Exa results) as described in the 'Research & Fact-Checking' section of SKILL.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the interpolation logic.
Capability inventory: The skill primarily performs text generation and formatting; it does not have file-write or general code execution capabilities.
Sanitization: No evidence of sanitization or filtering of external research content before processing.
[EXTERNAL_DOWNLOADS]: The skill references and encourages the installation of external software and configuration files from the 'inference-sh' organization.
Evidence: References installation instructions at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md and links to https://inference.sh.
Evidence: Recommends installing the belt-sh/cli and inference-sh/skills packages via npx.
[COMMAND_EXECUTION]: The skill provides numerous examples for executing shell commands via the belt CLI tool for authentication and data retrieval. The allowed-tools frontmatter correctly limits the scope of the Bash tool to only commands prefixed with belt, reducing the risk of arbitrary command execution.

press-release-writing