The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/gerrit_stream_events.py script executes the system ssh command via subprocess.Popen to connect to Gerrit's real-time event stream. This is a documented core feature and uses standard parameters for automation.
[CREDENTIALS_UNSAFE]: The skill manages sensitive data including Gerrit HTTP credentials and SSH private keys. It implements a secure pattern by loading these from non-distributed configuration files (gerrit_config.json) and environment variables, and the documentation correctly advises users to prevent these files from being committed to version control.
[DATA_EXFILTRATION]: The skill includes a feature to POST event data (such as project and change details) to a user-defined HTTP hook URL. This represents an intentional integration mechanism, and the documentation includes security warnings regarding its use over non-TLS connections.
[PROMPT_INJECTION]: The skill introduces an indirect prompt injection surface as it ingests untrusted data, such as commit messages and review comments, from the Gerrit server.
Ingestion points: REST API responses in gerrit_api.py and SSH stream output in gerrit_stream_events.py.
Boundary markers: Not explicitly defined in the script output; the agent consumes the raw or summarized content.
Capability inventory: Includes execution of ssh commands, writing to local files, and sending data to HTTP hooks.
Sanitization: Relies on standard JSON parsing for technical safety but does not perform content-level sanitization for natural language instructions that might influence the agent.

gerrit-api