The Agent Skills Directory

[NO_CODE]: The skill consists entirely of a markdown configuration file (SKILL.md) and lacks any executable scripts, binaries, or implementation logic.
[REMOTE_CODE_EXECUTION]: The documentation explicitly claims the capability to "Auto-install and configure" external MCP tools from a discovery pool of 3000+ items, which involves the execution of third-party code from potentially unverified sources.
[CREDENTIALS_UNSAFE]: The skill describes "API key management" and "Auto API key setup" as core features, indicating that the agent is expected to handle and potentially store sensitive authentication tokens.
[DATA_EXFILTRATION]: The orchestration of multiple unverified tools alongside managed API keys creates a risk where sensitive configuration data or secrets could be accessed or transmitted by third-party extensions.
[PROMPT_INJECTION]: The process of "discovering" external tools introduces an indirect prompt injection surface where malicious tool metadata or instructions could influence agent behavior.
Ingestion points: External MCP tool discovery process described in SKILL.md.
Boundary markers: None identified.
Capability inventory: Tool installation, configuration, and multi-tool workflow execution.
Sanitization: No verification or sanitization mechanisms are specified for the discovered tools.

mcp-orchestrator