# Rath-Team/OpenRath

An open-source, PyTorch-like runtime for dynamic multi-agent and multi-session workflows.

Repository: https://github.com/Rath-Team/OpenRath
Canonical: https://ross.abutalabs.com/products/openrath
Homepage: https://www.openrath.com/
Language: Python
License: BSD-3-Clause
License Family: permissive
Topics: agent-framework, agentic-ai, ai-agents, anthropic, llm, memory, model-context-protocol, multi-agent, multi-agent-systems, open-source, openai, provenance, python, runtime-state, sandbox, session-graph, session-state, workflow-orchestration, lllm-agent
Last push: 2026-07-31T05:04:23+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 95, release rhythm 95, longevity 8
- inputs: {"age_days": 121, "days_push": 33, "days_rel": 33, "gap_med": 6.0, "n_releases_24m": 7}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1097, forks 50 (observed 2026-08-28T04:03:34.509965+00:00)

## What it is
OpenRath is a PyTorch-like Python framework for building dynamic multi-agent and multi-session workflows. It models runtime state as explicit composable objects (Session, Sandbox, Memory, Tool, Agent, Workflow, Selector) with traceable lineage and sandboxed tool execution.

## Use cases
- build multi-agent systems with branching sessions
- orchestrate LLM agents with durable memory across runs
- run agent tools in sandboxed execution environments
- route between workflows dynamically with an LLM-backed selector
- audit and resume long-running agent tasks with full lineage
- compose agents and workflows like software modules
- build agents using OpenAI or Anthropic models

## When to choose
- you need many collaborating agents with branchable, inspectable session state
- you want durable memory and sandboxed tool execution in one runtime
- you prefer a composable, PyTorch-style API for agent workflows
- you need provenance and auditability of every tool call and state change

## When to avoid
- you only need a simple single-agent chat loop
- you want a mature, battle-tested framework with a large ecosystem
- you are not working in Python
- you need a no-code or UI-driven agent builder

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, workflow-automation, mcp, llm-inference, sdk
- domain: large-language-models, developer-tools
- platform: python, cross-platform
- tags: multi-agent, session-graph, sandboxed-execution, agent-memory, torch-like-api, workflow-orchestration, provenance, openai, anthropic, ai-agents, automation

## Member repositories
- Rath-Team/OpenRath (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:34.509965+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:46:23.231285+00:00, confidence not recorded.
  - readme: https://github.com/Rath-Team/OpenRath (fetched 2026-08-28T04:03:34.509965+00:00, sha 2cd1bce9e7ac)
  - homepage: https://www.openrath.com/ (fetched 2026-08-29T12:49:57.681994+00:00, sha 911cb45aae30)
  - site_page: https://docs.openrath.com/ (fetched 2026-08-29T12:49:57.684853+00:00, sha 299e18588cc7)
  - registry_pypi: https://pypi.org/pypi/openrath/json (fetched 2026-08-29T12:49:57.686637+00:00, sha 93510ca67d6e)
- Data as of 2026-08-30T08:39:29.467469+00:00.