no-materials/hird

Overview

A typed language for building LLM agent systems on the BEAM virtual machine. It tracks effect rows for tool calls, provides auditable tool effects, typed actors, and OTP supervision, and exposes agent queries over MCP.

Best for

Best for
Developers building reliable, auditable LLM agent systems on BEAM who value type safety and supervision.

Use cases

• Building type-safe agent workflows with auditable tool side effects
• Implementing supervised actor systems for LLM-driven processes
• Exposing agent state and capabilities to external tools via MCP

How to use

Tested with

Claude Code, Cursor

Notes

A typed language for building LLM agent systems on the BEAM virtual machine. It tracks effect rows for tool calls, provides auditable tool effects, typed actors, and OTP supervision, and exposes agent queries over MCP.

1 stars on GitHub. Last updated 2026-09-07. Licensed Apache-2.0.

Use cases

• Building type-safe agent workflows with auditable tool side effects
• Implementing supervised actor systems for LLM-driven processes
• Exposing agent state and capabilities to external tools via MCP

Pros

• Effect-row tracking makes tool side effects explicit and checkable
• Leverages OTP supervision for robust actor lifecycle management
• Typed actors reduce runtime errors in concurrent agent logic

Cons

• Very early stage with only 1 star and minimal adoption
• Requires BEAM runtime, limiting deployment to Erlang/Elixir ecosystems
• Rust implementation adds a language boundary for BEAM developers

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.