The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes an installation command that downloads a script from github.com/perplexityai and pipes it directly into the shell (curl -fsSL ... | sh). This is the official installation method provided by the tool's vendor.
[EXTERNAL_DOWNLOADS]: The skill fetches the pplx binary, installation scripts, and updates from the vendor's official GitHub releases and talks to api.perplexity.ai for search queries.
[COMMAND_EXECUTION]: The skill relies on executing the pplx CLI tool via the shell to perform web searches, retrieve content snippets, and manage authentication.
[CREDENTIALS_UNSAFE]: The skill documents the location of stored API credentials at ~/.config/perplexity/credentials.json and supports the use of the PERPLEXITY_API_KEY environment variable for non-interactive authentication.
[PROMPT_INJECTION]: The skill fetches external content from the web which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Untrusted content is ingested via the pplx content snippets command which retrieves text from user-provided URLs.
Boundary markers: The content is returned in a structured JSON format, but the skill does not specify explicit delimiters or instructions for the agent to ignore instructions embedded within the retrieved snippets.
Capability inventory: The skill can execute shell commands (pplx) and access the file system to save search results.
Sanitization: No specific sanitization or filtering of the retrieved web content is mentioned beyond the CLI's internal truncation and token budgeting features.

pplx-cli