The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the execution of multiple shell commands to build the plugin, run a background relay, and perform transport probes.
Evidence: Commands such as bun packages/mcp/scripts/probe-transports.ts, bun --filter figma-mcp build, and bun packages/mcp/bin/index.mjs socket are used throughout the verification phases.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process Figma URLs provided by a user, which serves as a potential vector for indirect instructions if the URL or its metadata contains malicious payloads.
Ingestion points: The skill explicitly requests a 'layer URL' from the user via AskUserQuestion in Phase 0 to derive file keys and node IDs.
Boundary markers: Absent. There are no explicit instructions for the agent to treat the user-supplied URL as untrusted or to ignore embedded instructions within the parsed content.
Capability inventory: The skill has the ability to execute shell commands (bun), initiate network connections (REST and WebSocket), and read/write to the local filesystem for building and logging.
Sanitization: The skill mentions using a parseFigmaUrl function, but there is no evidence of robust sanitization or strict schema validation for the input data beyond extraction.

seed-verify-figma-mcp-transports