The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses lokei setup to perform high-risk system modifications, including configuring a local DNS resolver and establishing port forwarding rules (pfctl on macOS or iptables on Linux) for ports 80 and 443.\n- [COMMAND_EXECUTION]: The lokei service install command allows for the installation of persistent background services as launchd or systemd daemons, which maintains the tool's execution across system restarts.\n- [COMMAND_EXECUTION]: During the setup process, the skill triggers certificate generation and trust operations that modify the OS keychain, representing a privilege escalation vector that requires administrative access.\n- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download and install the lokei package globally from the NPM registry if it is missing from the environment.\n- [DATA_EXFILTRATION]: Through the lokei share and lokei run --expose commands, the skill can create public tunnels that route local traffic through relay.lokei.dev, potentially exposing local development data to the public internet.\n- [PROMPT_INJECTION]: The skill provides access to network logs and request metadata, which constitutes an indirect prompt injection surface if malicious external data is processed.\n
Ingestion points: Network traffic logs and request details accessed via lokei logs or lokei inspect.\n
Boundary markers: None. The skill does not implement delimiters or 'ignore' instructions for log content.\n
Capability inventory: The agent has access to the Bash tool, allowing it to execute commands if influenced by malicious log data.\n
Sanitization: No content filtering or escaping is described for the log viewing functionality.

lokei