brainstorming
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a 'Visual Companion' that logs user interactions in a browser to a local file (
state_dir/events) which the agent then reads. - Ingestion points: The agent reads interaction data from the
$STATE_DIR/eventsfile as part of its brainstorming workflow. - Capability inventory: The agent has the ability to execute shell commands, write files, and invoke other planning skills.
- Boundary markers: The instructions provide context on how to interpret the JSON events but do not include explicit security delimiters or 'ignore' instructions for the data read from the events file.
- Sanitization: No explicit sanitization or validation of the event data is performed before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill utilizes shell scripts (
start-server.sh,stop-server.sh) and a Node.js server (server.cjs) to provide its visual companion functionality. These scripts manage local processes, track PIDs, and bind to local network ports to serve content and receive events.
Audit Metadata