ai-netcafe
Verified · 21 days agoTables and ledgers checked by arithmetic, not by a model. 24 tools. MCP 2026-07-28 ready.
MCP server for AWS S3 with fine-grained tool control and S3-compatible endpoint support.
claude mcp add s3 -- npx -y [email protected]
An S3 server from PulseMCP (the MCP directory company) living in the experimental/ folder of their mcp-servers monorepo: nine tools spanning bucket and object CRUD, published to npm as s3-aws-mcp-server, TypeScript over stdio with standard AWS credential resolution. Its differentiator is permissioning rather than feature count — readonly/readwrite tool groups, per-tool enable/disable env vars, and a single-bucket mode that scopes every operation to one bucket and hides bucket-level tools entirely, which is exactly the containment story you want before handing an agent delete_object. It also speaks to S3-compatible endpoints (MinIO, LocalStack), and the monorepo is actively maintained (72 stars, pushed the day of review), though only 2 contributors and ~51 weekly downloads, and the 'experimental' label is the maintainers' own signal to temper expectations. For teams giving an agent scoped S3 access who care more about blast-radius control than exhaustive API coverage.
Tables and ledgers checked by arithmetic, not by a model. 24 tools. MCP 2026-07-28 ready.
Puter MCP enables AI tools to interact with Puter: manage files, websites, workers, and more
MCP server for filesystem access
MCP server for terminal commands, file operations, and process management
Local-first knowledge management with bi-directional LLM sync via Markdown files.
Markdown knowledge base as agent memory. Runs against the notes directory it is started in.