fill-workday-timesheet

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the chrome-cdp CLI to perform complex browser automation, including simulating mouse events and form interactions.
  • [COMMAND_EXECUTION]: Accesses local user configuration by sourcing a file from ~/.config/harness-configs/fill-workday-timesheet/config to retrieve settings such as project names and default hours.
  • [DYNAMIC_EXECUTION]: Utilizes browser-side JavaScript evaluation (eval) to programmatically retrieve element IDs from the DOM when standard selectors are insufficient.
  • [DYNAMIC_EXECUTION]: Executes external 'recipes' (pre-defined automation scripts) stored in ~/.config/chrome-cdp/recipes/ at runtime.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes live data from the Workday web interface (e.g., week headings and hour totals) to determine its execution flow, creating a surface for indirect injection if page content is manipulated.
  • Ingestion points: External UI text is captured through chrome-cdp snap and chrome-cdp find operations.
  • Boundary markers: No explicit delimiters are used to isolate ingested browser data from the automation logic.
  • Capability inventory: The skill possesses extensive control over the user's browser session via chrome-cdp, including the ability to write data and execute scripts.
  • Sanitization: Ingested UI content is not sanitized or validated before influencing subsequent browser commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 08:52 AM
Security Audit — agent-trust-hub — fill-workday-timesheet