essential-mcp-servers-reference
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides documentation and commands to download various MCP server packages from NPM using
npx -y. These include official packages under the@modelcontextprotocolscope as well as well-known community and vendor packages (e.g.,@kubernetes/mcp-server,@alpacahq/mcp-server,@sentry/mcp). It also references hosted MCP endpoints from established providers such as Supabase, Neon, Cloudflare, Vercel, and Sentry. - [COMMAND_EXECUTION]: The skill documents the use of shell commands (
claude mcp add,codex mcp add) to configure AI agents with external capabilities. These commands are standard for the MCP ecosystem and the skill includes significant warnings about treating external servers as untrusted code and enforcing least-privilege principles. - [INDIRECT_PROMPT_INJECTION]: The skill describes tools that ingest data from external sources (GitHub PRs, Slack messages, Notion documents). This represents a standard vulnerability surface where malicious external data could attempt to influence agent behavior.
- Ingestion points: External APIs (GitHub, Slack, Jira, Notion, Google Drive) and local filesystem operations as defined in
SKILL.md. - Boundary markers: Not explicitly defined in the reference guide; the skill relies on the agent's native handling and the provided security advice (e.g., read-only access).
- Capability inventory: The guide lists tools with significant capabilities including filesystem access, database write access, and financial transaction execution (Stripe, Alpaca).
- Sanitization: The skill recommends manual sanitization through read-only database configurations and fine-grained API token scoping (e.g., GitHub Personal Access Tokens).
Audit Metadata