The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's architecture explicitly instructs the bridge to dispatch tasks to Claude Code using the --dangerously-skip-permissions flag. This configuration bypasses the human-in-the-loop security model, granting the agent the ability to execute shell commands and modify files autonomously without requesting user approval.
[REMOTE_CODE_EXECUTION]: The --dispatch functionality allows for the execution of arbitrary tasks described in natural language. When combined with the removal of permission prompts, this creates a high-risk environment where complex, potentially unsafe operations are executed by a secondary agent without oversight.
[COMMAND_EXECUTION]: Troubleshooting instructions suggest the execution of a local ./install.sh script and the creation of symbolic links to the kc.js executable to enable the ccc command.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of the @anthropic-ai/claude-code package from the official npm registry, which is a well-known and trusted source.
[DATA_EXFILTRATION]: The skill utilizes the CC_OPENCLAW_URL environment variable to forward event data. While it defaults to a local address, this mechanism is designed to transmit agent session data and results to a specified endpoint, which could be configured to an external server.

openclaw-ccc-bridge