wecomcli-sheet

Warn

Audited by Gen Agent Trust Hub on Nov 3, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/recalc.py contains functionality to compile an embedded C source shim at runtime using gcc. This compiled library is then injected into the soffice (LibreOffice) process using the LD_PRELOAD environment variable. This mechanism is intended to shim socket-related system calls to enable communication in restricted sandbox environments where AF_UNIX sockets are blocked.\n- [COMMAND_EXECUTION]: The skill performs shell command execution via subprocess.run in scripts/recalc.py to invoke gcc, soffice, and timeout utilities (timeout/gtimeout).\n- [INDIRECT_PROMPT_INJECTION]: The skill reads and analyzes data from external spreadsheet files, creating a vulnerability surface where malicious instructions embedded in cells could be processed by the agent.\n
  • Ingestion points: Untrusted data is ingested from local .xlsx, .xlsm, .xls, .csv, and .tsv files via scripts/collect.py.\n
  • Boundary markers: While SKILL.md includes safety instructions for the agent to ignore commands within processed data, there are no programmatic boundary markers or delimiters enforced on the data passed to the prompt.\n
  • Capability inventory: The skill has the capability to write to local and online files, execute shell commands, and modify spreadsheet structures.\n
  • Sanitization: No programmatic sanitization or filtering is performed on the cell values before they are analyzed by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 3, 2026, 02:48 AM