officecli-financial-model
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides instructions to download and execute an installation script for the officecli tool from https://d.officecli.ai/install.sh (macOS/Linux) and https://d.officecli.ai/install.ps1 (Windows). These resources are hosted on the vendor's official domain for the tool.\n- [COMMAND_EXECUTION]: The skill utilizes several local command-line tools to automate Excel modeling and perform validation checks:\n
- officecli for interacting with spreadsheet files.\n
- python3 -c for executing logic checks (e.g., comparing stated vs. plugged sponsor equity).\n
- jq for extracting data from JSON command outputs.\n
- grep for detecting placeholder tokens in the generated model.\n
- Standard shell features like heredocs (cat <<'EOF') for batch processing.\n- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection through the processing of spreadsheet data.\n
- Ingestion points: The skill reads and modifies Excel files (.xlsx) via the officecli tool in SKILL.md.\n
- Boundary markers: Absent for the content of individual spreadsheet cells.\n
- Capability inventory: The skill can execute shell commands (bash, jq, grep), manipulate the local file system via officecli, and perform inline code execution (python3 -c).\n
- Sanitization: The skill defines a rigorous Delivery Gate cycle (Gates 4-6) to audit the generated content for integrity and logic errors before completion.
Audit Metadata