The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): The skill explicitly commands the execution of a remote script from an untrusted source via curl -fsSL https://cli.inference.sh | sh. This is a severe security risk as the remote script can be modified by the provider to execute arbitrary malicious code on the host system.
[EXTERNAL_DOWNLOADS] (HIGH): The skill directs the installation of multiple external modules using npx skills add inference-sh/skills@.... These sources are not among the trusted GitHub organizations or repositories and have not been verified for safety.
[COMMAND_EXECUTION] (MEDIUM): The skill requires the use of the infsh CLI tool, which executes bash commands. This tool is granted broad access (infsh *) to perform operations like searching the web and posting to social media, increasing the potential impact of a compromise.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill is susceptible to indirect prompt injection because it ingests data from untrusted external sources. 1. Ingestion points: Untrusted data enters the agent context via search results from tavily/search-assistant and exa/search. 2. Boundary markers: Absent; there are no instructions or delimiters to isolate search results from the agent's instructions. 3. Capability inventory: The agent has capabilities to execute bash commands, perform network searches, and create social media posts. 4. Sanitization: Absent; the skill does not include any validation or escaping of the content retrieved from the web before processing it.

newsletter-curation