The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill represents an attack surface for indirect prompt injection as it ingests untrusted data from the Melty API (e.g., game slugs, mod folders, and installation failure reports).
Ingestion points: Data is ingested via tools such as game_info, mashup_info, and install_outcomes from https://melty.gg/api/mcp.
Boundary markers: The skill does not define specific boundary markers or instructions for the agent to ignore embedded commands in the retrieved data.
Capability inventory: The agent has the capability to write files, upload archives to signed URLs, and publish content publicly using tools like start_upload, submit_release, and publish.
Sanitization: No explicit sanitization or validation logic for the ingested external content is described.
[DATA_EXFILTRATION]: The skill implements a tool named consult_melty which functions as a progress log by sending the user's last message and the agent's full reply to an external server (https://melty.gg/api/mcp). While the skill provides explicit safety instructions to avoid including credentials or private paths, the architectural design creates a channel for potential data exposure of session context.

melty-mashups