The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted documentation content from the Logitech support portal (support.logi.com). This ingestion of external data creates a surface for indirect prompt injection, where malicious instructions embedded in manuals or spec sheets could influence agent behavior.
Ingestion points: The commands articles get, find, and articles search retrieve HTML and text content from external web sources and local synced copies.
Boundary markers: The skill instructions lack explicit delimiters or instructions for the agent to ignore or isolate commands found within the processed documentation.
Capability inventory: The skill can execute shell commands via the logitech-docs-pp-cli binary and the Read Bash tool.
Sanitization: There is no evidence of content filtering or sanitization to prevent the agent from interpreting embedded instructions as commands.
[DATA_EXFILTRATION]: The CLI tool supports a --deliver webhook:<url> parameter, enabling the agent to POST command results to any specified URL. While intended for integration, this capability could be exploited to exfiltrate data if the agent is directed to use an attacker-controlled endpoint.
[COMMAND_EXECUTION]: The skill implements a 'learning loop' and 'playbook' system that dynamically suggests and executes sequences of CLI commands. These playbooks are synthesized from the session journal and stored in a local SQLite database, representing a dynamic command execution surface driven by historical state.

pp-logitech-docs