Ae1
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md - `scripts/find-sessions.sh`: discover sessions.
Security audit
Security checks for vulnerabilities and agentic risk
This tmux skill is coherent, but it gives agents broad control over live terminal sessions and can expose pane contents without enough safety guidance.
Review before installing. Use this only when you are comfortable letting the agent inspect and type into tmux panes. Confirm exact session/window/pane targets before sending input, avoid C-c/C-d or kill-session unless explicitly intended, and do not use the wait helper on panes that may contain secrets unless you accept that timeout logs may include pane output.
Referenced artifact was not completely inspected
- `scripts/find-sessions.sh`: discover sessions.
The documentation shows sending control keys like C-c and C-d to live panes without warning that these inputs commonly interrupt, abort, log out of, or terminate running programs and shells. Because this skill is meant for interacting with existing tmux panes, using such keys against the wrong pane can cause unintended service interruption, data loss, or premature termination of active tasks.
The skill documents tmux kill-session -t worker alongside other session commands without any warning that it will immediately terminate the tmux session and any processes attached to it. In a skill specifically designed to control live interactive sessions, this increases the chance an agent or user will apply the command to the wrong target and disrupt active work or kill long-running jobs.
On timeout, the script prints the last captured tmux pane contents to stderr, which can expose sensitive data such as credentials, tokens, command output, or other secrets present in the pane history. In this skill's context, tmux is explicitly used to control interactive CLIs, making it likely to handle confidential terminal content and increasing the risk of unintended disclosure through logs or higher-level agent error surfaces.
No suspicious patterns detected.