Releases·Tool Release
Developer releases AURA Harness, an offline agent harness that wraps existing agent loops
A developer introduced AURA Harness, a local-first agent harness designed to conform to existing agent runtimes rather than requiring users to rebuild their loops around a framework. It acts as a thin membrane around loops already running in plain Python, Ollama, or LangGraph, recording and optionally gating what crosses the boundary. The approach matters to practitioners frustrated that most agent frameworks force runtime rewrites.
1 engagement1 sourceMon, Aug 31, 2026, 09:06 AM
AI Overview
- AURA Harness works offline using agent_ref plus session IDs, with no cloud identity required
- Includes an Ollama integration (integrations/ollama/llama_loop.py) using stdlib HTTP calls
- Author reports testing many harness models and frameworks over the past month before building it
- Author argues harnesses should record and gate boundary crossings while users keep their own runtime
Entities
Related
Social·Community
What your agent framework hides, and what you learn by building the harness yourself in plain Python
A four-hour workshop titled Harness Engineering will teach attendees to build an AI data analyst without any agent framework, in plain Python. Led by Bruno Gonçalves, founder of Data For Science, it runs Saturday, Oct 31, from 7 PM to 11 PM IST. The pitch: writing a harness yourself reveals what frameworks hide — which layer retries, which tool result blows up the context, why a guardrail fails.
A four-hour workshop titled Harness Engineering will teach attendees to build an AI data analyst without any agent framework, in plain Python. Led by Bruno Gonçalves, founder of Data For Science, it runs Saturday, Oct 31, from 7 PM to 11 PM IST. The pitch: writing a harness yourself reveals what frameworks hide — which layer retries, which tool result blows up the context, why a guardrail fails.
1 engagement1 sourceredditTue, Oct 13, 2026, 09:21 AM
Products·building
Open-source self-healing security pipeline for LangGraph + Ollama multi-agent workflows
An open-source pipeline that sits around LangGraph multi-agent workflows running Llama 3.1 via Ollama locally. It addresses parser differentials (malformed markdown or injected keys in JSON tool calls) and state desynchronization from async processing, using a canonical schema guardrail and self-healing validation. Intended for developers building local multi-agent systems.
An open-source pipeline that sits around LangGraph multi-agent workflows running Llama 3.1 via Ollama locally. It addresses parser differentials (malformed markdown or injected keys in JSON tool calls) and state desynchronization from async processing, using a canonical schema guardrail and self-healing validation. Intended for developers building local multi-agent systems.
1 engagement1 sourceredditFri, Oct 2, 2026, 02:13 PM
Releases·Tool Release
Developer launches open-source local-first workflow automation platform with Ollama integration
A developer released v0.11.0 of an open-source workflow automation platform that treats Ollama as a first-class provider, enabling local LLMs to power deterministic automation pipelines. The platform supports agent semantic memory, document RAG, visual workflow builder, conditional branching, human-in-the-loop approval nodes, and multi-agent orchestration.
A developer released v0.11.0 of an open-source workflow automation platform that treats Ollama as a first-class provider, enabling local LLMs to power deterministic automation pipelines. The platform supports agent semantic memory, document RAG, visual workflow builder, conditional branching, human-in-the-loop approval nodes, and multi-agent orchestration.
4 engagement1 sourceredditMon, Jul 27, 2026, 07:03 AM
Developments·Paper
Arxiv paper analyzes architectural convergence in three LLM agent harnesses
A new arxiv paper examines three open coding-agent harnesses—LangChain's deepagents, Earendil's pi, and DeepSeek's dsh—highlighting architectural convergence despite differing design philosophies. The study argues that the harness layer, not the underlying model, is the primary constraint on agent behavior, offering insights for practitioners building autonomous agents.
A new arxiv paper examines three open coding-agent harnesses—LangChain's deepagents, Earendil's pi, and DeepSeek's dsh—highlighting architectural convergence despite differing design philosophies. The study argues that the harness layer, not the underlying model, is the primary constraint on agent behavior, offering insights for practitioners building autonomous agents.
0 engagement1 sourcearxivTue, Aug 25, 2026, 01:26 AM
Releases·Tool Release
Author releases OnceMesh, an open source system for safely reusing exact LLM agent work
OnceMesh is an open source system that identifies and reuses exact LLM agent computations, storing outputs as content-addressed artifacts after checking authorization, freshness, provenance, and integrity. It targets repeated expensive work like downloads, parsing, and deterministic tool calls, and is not a semantic prompt cache.
OnceMesh is an open source system that identifies and reuses exact LLM agent computations, storing outputs as content-addressed artifacts after checking authorization, freshness, provenance, and integrity. It targets repeated expensive work like downloads, parsing, and deterministic tool calls, and is not a semantic prompt cache.
27 engagement1 sourceredditTue, Aug 25, 2026, 03:39 PM