AgentOps addresses a real pain point in AI coding workflows: how to persist knowledge and orchestrate multiple AI agents across different vendors locally, without cloud dependencies or telemetry. Its standout feature is the /council command, which runs multiple agents independently to review the same evidence and produce a consolidated verdict — a level of multi-agent consensus rarely seen in AI tooling.
local cross-vendor orchestration for ai coding agents
AgentOps is a Go-based local operating layer designed to unify AI coding agents from various vendors under a single control plane. It uses the ao CLI as the primary interface to manage agents, their workflows, and persistent memory.
Under the hood, it maintains a local .agents/ directory that acts as a persistent memory corpus for session-to-session knowledge. This allows agent context and outputs to survive across sessions and tooling boundaries, sidestepping the stateless nature common in many AI agent setups.
The architecture applies the DevOps “Three Ways”: flow, feedback, and continual learning — principles rarely formalized in AI agent orchestration. AgentOps introduces validation gates such as /pre-mortem to catch issues upfront, /vibe for quality checks, and /council for multi-agent review and consensus.
Composable flows like /research, /implement, and /validation organize agent tasks into distinct phases, enabling structured workflows that mirror real-world development processes.
The stack is pure Go, focusing on zero telemetry and zero cloud dependencies. All state is stored locally and explicitly git-ignored to avoid accidental commits.
multi-agent consensus and local persistent memory as technical differentiators
What sets AgentOps apart is its multi-agent evaluation approach, especially visible in the /council command. This orchestrates multiple AI agents (for example, Claude and Codex) to independently analyze the same inputs and produce individual assessments. These assessments are then consolidated into an auditable verdict.
This design enforces a clear separation between replaceable agents and the operating layer that manages state and context. The local .agents/ memory corpus enables knowledge persistence that crosses vendor boundaries — a significant advantage for long-running or complex workflows.
The code is opinionated but pragmatic, reflecting real-world tradeoffs. The CLI is designed for developer experience, offering commands for bookkeeping, health checks, and terminal workflows that integrate naturally into repo-native contexts.
The zero telemetry and cloud-free design are deliberate tradeoffs to preserve privacy and control, but they imply that integrations with cloud-only services or remote model runtimes require explicit user choice.
Under the hood, the codebase leans on composable commands and structured workflows, which keeps complexity manageable despite the multi-agent orchestration. The modular design makes adding new agents or validation gates straightforward.
quick start with agentops
To get started, choose your runtime and follow the corresponding installation commands exactly:
Claude Code runtime:
claude plugin marketplace add boshu2/agentops
claude plugin install agentops@agentops-marketplace
Codex CLI on macOS, Linux, or WSL:
curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.sh | bash
Codex CLI on Windows PowerShell:
irm https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.ps1 | iex
OpenCode runtime:
curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-opencode.sh | bash
Other skills-compatible agents:
npx skills@latest add boshu2/agentops --cursor -g
After installation, restart your agent and use /quickstart in the agent chat to begin. Optionally, install the ao CLI for enhanced repo-native workflows.
For macOS:
brew tap boshu2/agentops https://github.com/boshu2/homebrew-agentops
brew install agentops
ao version
For Windows PowerShell:
irm https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-ao.ps1 | iex
ao version
The CLI unlocks features like local bookkeeping, retrieval, and health checks, integrating deeply with your repo environment.
verdict: who benefits from agentops
AgentOps is relevant for developers and teams working with multiple AI coding agents who need persistent, local memory and structured workflows across vendor boundaries.
Its multi-agent review system is a notable step towards more reliable AI-assisted development, especially in environments where auditability and local control are priorities.
The zero telemetry, zero cloud dependency approach suits privacy-conscious users but limits immediate cloud service integrations.
While the setup requires some familiarity with CLI tools and AI runtimes, the structured commands and validation gates help enforce best practices in AI coding workflows.
In practice, AgentOps is a solid choice if you want an opinionated, local operating layer that brings DevOps principles into AI agent orchestration, especially if you value multi-agent consensus and persistent context over purely cloud-based solutions.
Related Articles
- Inside agents: a granular multi-agent orchestration system with PluginEval quality assurance — Explore agents, a Python-based multi-agent orchestration repo featuring 184 AI agents, 78 plugins, and a three-layer Plu
- Agno: Building production-ready agentic software with minimal code — Agno provides a minimal, production-ready Python framework for scalable agentic software with per-user isolation and nat
→ GitHub Repo: boshu2/agentops ⭐ 328 · Go