The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and executes the @webmcp-bridge/local-mcp package from the NPM registry using npx within the ensure-links.sh script.
[COMMAND_EXECUTION]: The ensure-links.sh script dynamically constructs and executes a command string using uxc link to register the board-webmcp-cli tool. It utilizes a shell_join helper to handle argument escaping.
[DATA_EXFILTRATION]: The skill manages browser session data and persistence by creating and accessing a dedicated directory at ~/.uxc/webmcp-profile/board. This is used to maintain the state of the board bridge sessions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from a shared public board.
Ingestion points: Data enters the agent context through board-webmcp-cli nodes.list and board-webmcp-cli edges.list as documented in SKILL.md.
Boundary markers: No explicit delimiters or boundary instructions are used when processing the board state.
Capability inventory: The skill provides the ability to modify the board state and execute layout algorithms through board-webmcp-cli subcommands (upsert, apply).
Sanitization: No explicit sanitization or validation of the board data is performed before it is interpreted by the agent.

board-webmcp