The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install the unifi-pp-cli tool using npx from @mvanhorn/printing-press-library and go install from the vendor's GitHub repository. These are documented as standard installation procedures for the tool.
[COMMAND_EXECUTION]: The skill makes extensive use of the unifi-pp-cli binary to perform network operations, such as viewing topologies, auditing ports, and managing firewall rules.
[DYNAMIC_EXECUTION]: The skill implements a 'learning loop' that instructs the agent to execute 'trial commands' from candidates and 'playbook steps' retrieved from a local SQLite database at runtime. This allows for dynamic command generation and execution based on previous tool outputs.
[DATA_EXFILTRATION]: The CLI supports an output delivery mechanism via --deliver webhook:<url> which allows POSTing command results to arbitrary external URLs. Additionally, the feedback command can transmit local data to a remote endpoint if UNIFI_FEEDBACK_ENDPOINT is configured.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted data from network devices and a local learning store while maintaining the capability to execute shell commands.
Ingestion points: Output from unifi-pp-cli recall and various UniFi Network API responses (firewall rules, device names).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are provided for data processing.
Capability inventory: The skill has permission to use the Bash tool to execute the unifi-pp-cli binary.
Sanitization: No explicit sanitization or validation of external content is mentioned for the execution paths.

pp-unifi