The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @lessie/cli package globally using npm install -g and utilizes @lessie/mcp-server via npx. These resources are official components provided by the vendor, LessieAI.
[COMMAND_EXECUTION]: The skill's core functionality is built around executing bash commands. It uses the lessie CLI for searching people and organizations, performing contact enrichment, and managing authentication.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its web research features. The web_search and web_fetch tools ingest data from external, untrusted websites which could contain instructions aimed at manipulating the agent's output or behavior.
Ingestion points: Data retrieved from external URLs via web_search and web_fetch results.
Boundary markers: The instructions do not define specific delimiters to isolate untrusted web content from the system prompt.
Capability inventory: The agent can execute shell commands through the lessie CLI and access local configuration files like ~/.lessie/oauth.json.
Sanitization: There is no explicit logic in the skill to sanitize or filter out potential instructions within the fetched web content before it is processed by the agent.

lessie