The Agent Skills Directory

[DATA_EXFILTRATION]: The skill facilitates significant system information exposure through several scripts.
scripts/env.js and scripts/env.cjs access process.env and allow retrieving values via the --show-values flag. While the maskValue function attempts to redact keys containing 'password' or 'secret', it can be bypassed by custom environment variable names or simply by exposing non-masked but sensitive data like USER, HOME, and SHELL paths.
scripts/processes.js and scripts/processes.cjs list all running processes including their full command lines. This is a common source of credential leakage as tokens, API keys, and passwords are frequently passed as command-line arguments.
scripts/overview.js and scripts/overview.cjs expose sensitive system metadata including network interface MAC addresses, IP addresses, os.homedir(), process.cwd(), and process.execPath.
[COMMAND_EXECUTION]: The scripts/processes.js script utilizes execSync to run system commands (ps aux on Unix/macOS and tasklist on Windows). While the command strings are currently static, the use of shell execution for process monitoring presents an attack surface if argument parsing were to be modified.
[DYNAMIC_EXECUTION]: The use of execSync in scripts/processes.cjs and scripts/processes.js for process listing qualifies as dynamic execution of system binaries.
[INDIRECT_PROMPT_INJECTION]: The scripts/health.js script introduces an indirect prompt injection surface.
Ingestion points: Reads and parses the JSON body of responses from arbitrary remote URLs via fetch().
Boundary markers: None present; the script returns raw JSON bodies from remote services back to the agent context.
Capability inventory: The skill possesses capabilities for file system inspection (overview.js), environment variable access (env.js), and further network requests (health.js).
Sanitization: No sanitization or validation is performed on the remote content before it is returned to the agent.

system-info