Omnigent is an open-source Python-based agent harness and execution orchestrator developed by the Omnigent AI team. It is designed to coordinate modular autonomous agents by managing runtime execution, sandboxed hosts, tool interfaces, and artifact persistence across developer and enterprise environments. In daily operations, Omnigent manages structured agent workflows through capability-driven modules. It provides built-in support for execution sandboxes via OpenShell, telemetry with OpenTelemetry GenAI semantic conventions, and structured message routing via Protocol Buffers. Developers configure agent runs using standard Python packaging conventions, enabling selective capability loading for specialized research, testing, or storage backends. What distinguishes Omnigent from standalone interactive CLI coding agents is its focus on harness-level orchestration and modularity. Instead of coupling the prompt interface directly to local execution, it decouples the agent runtime, allowing teams to execute agent commands safely inside managed sandboxes, log detailed traces across tool calls, and persist generated artifacts directly to S3 or relational database backends. Omnigent is open source on GitHub under standard open-source licensing, featuring an active continuous integration pipeline, extensive unit/e2e test suites, and modern dependency management with uv.
Tags: agent harness, ai agent, cli, coding agent, developer tools, OpenSource
openshell, s3, nimble, hindsight, and tracing. - OpenShell sandbox support — Provides isolated execution environments and managed hosts for running agent commands without risking the host machine. - OpenTelemetry observability — Integrates tracing and telemetry following standardized GenAI semantic conventions for tracking tool executions and agent latency. - S3 and D1 artifact storage — Built-in drivers to store, index, and retrieve workspace artifacts and run outputs across S3 buckets and SQLite/D1 databases. - Protobuf-based routing — Uses Protocol Buffer schemas to handle structured message passing and internal routing between agent components.Omnigent is built for platform engineers, AI infrastructure teams, and Python developers building custom agent runtimes that require sandboxing, telemetry, and modular tool harnesses. It is not intended for non-technical users looking for a plug-and-play desktop chat application.
uv or pip: bash # Basic installation uv pip install omnigent # Installation with OpenShell sandboxing, S3 storage, and tracing uv pip install "omnigent[openshell,s3,tracing]"bash # Set model provider and telemetry environment variables export OPENAI_API_KEY="sk-..." export OTEL_EXPORTER_OTLP_ENDPOINT="http://localhost:4317" # Verify the package installation python -c "import omnigent; print(omnigent.__name__)"bash # Run an agent workflow via Python or CLI runtime python -m omnigent.runtime --sandbox openshell --task "Analyze repo structure and generate migration plan"Part of CLI Coding Agents
Categories: AI Coding Agents, Code Assistants, Developer Tools