terminal-session-control
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from terminal sessions, creating a potential surface for indirect prompt injection. \n
- Ingestion points: The skill utilizes
herdr agent readandcmux read-screeninSKILL.mdto ingest output from active terminal panes and background agents.\n - Boundary markers: The instructions include a specific mitigation strategy to "Preserve literal text, quoting, and multiline boundaries; do not use shell interpolation to construct untrusted payloads."\n
- Capability inventory: The skill has the ability to send input back to the terminal using
herdr agent promptandcmux sendas described inSKILL.md.\n - Sanitization: The skill provides clear instructions to avoid shell interpolation, which helps prevent the agent from accidentally executing commands embedded in the terminal output.\n- [COMMAND_EXECUTION]: The skill relies on executing local command-line interface tools (
herdrandcmux) to manage terminal sessions. This is the core functionality and is handled by advising the agent to consult the installed tool's help documentation for the correct syntax.
Audit Metadata