The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates and executes Python scripts to perform network diagnostics and management tasks via the UniFi and AT&T Gateway APIs. These scripts are created at runtime based on user requests.
[CREDENTIALS_UNSAFE]: The skill requires administrative credentials for the UniFi controller and AT&T Gateway access codes, storing configuration data in a local directory (~/.bishop/config.json). While it encourages use of an encrypted store via creds.py, the initial configuration involves plaintext input.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the network environment, including client hostnames, SSID names, and IDS/IPS event logs (ingested via stat/sta and stat/ips/events in SKILL.md). There are no explicit boundary markers or sanitization steps before this data is used to generate reports or trigger capabilities like blocking clients or restarting hardware.
[DATA_EXPOSURE]: The skill's implementation guide instructs the agent to disable SSL certificate verification (verify=False in Python requests) to accommodate self-signed certificates on local network hardware, which could expose communication to local man-in-the-middle interception.
[DATA_EXFILTRATION]: The alerter.py module is designed to send network health reports and critical alerts to an external email address provided in the configuration.

bishop