The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses the !command syntax in the SKILL.md file to execute shell commands when the skill is loaded. It specifically runs version checks for Node.js and Python, and performs a length check on the PERPLEXITY_API_KEY environment variable.
[CREDENTIALS_UNSAFE]: Step 2 of the diagnostic instructions captures the first five characters of the PERPLEXITY_API_KEY variable and writes them to a plain-text file (summary.txt). This constitutes a partial leak of sensitive credentials.
[COMMAND_EXECUTION]: The instructions prompt the creation of a Bash script that executes multiple system tools, including uname, npm, pip, dig, and tar. It also reads environment variables and the contents of .env files.
[EXTERNAL_DOWNLOADS]: The skill uses curl to make network requests to api.perplexity.ai to verify API health and model availability. These requests target a well-known service relevant to the skill's primary purpose.
[DATA_EXPOSURE]: The diagnostic script reads the local .env file and system environment variables. Although it attempts to redact the values using sed, reading these sensitive files into a debug bundle increases the risk of data exposure if the redaction logic fails or is bypassed.

perplexity-debug-bundle