skills/daxaur/openpaw/c-clipboard/Gen Agent Trust Hub

c-clipboard

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands using standard system utilities such as pbcopy, pbpaste, xclip, and xsel to interact with the host operating system's clipboard.
  • [DATA_EXFILTRATION]: Provides the capability to read the system clipboard, which often contains sensitive data (e.g., passwords, keys, private text). The skill attempts to mitigate this with a guideline instructing the agent not to display clipboard contents unless specifically requested by the user.
  • [PROMPT_INJECTION]:
  • Ingestion points: System clipboard data read via pbpaste or xclip -o serves as an entry point for untrusted external data.
  • Boundary markers: None present to distinguish between user commands and clipboard content.
  • Capability inventory: Includes reading/writing files, piping to other shell transformations (tr, sort, wc), and echoing text.
  • Sanitization: No explicit sanitization or escaping of clipboard content is performed before processing or display. This creates a surface for indirect prompt injection if the clipboard contains instructions intended to manipulate the agent's behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 08:44 AM
Security Audit — agent-trust-hub — c-clipboard