CLI for A2A agent discovery, task control, and streaming results
Fetch AgentCards, create tasks, and receive deterministic NDJSON or streaming results over REST or WebSockets. a2acli, from G. Hussain Chinoy, bridges humans and AI agents for interactive or automated workflows. It offers an interactive TUI for manual exploration and a mock local agent server for development, plus machine-friendly JSON outputs. Developers building agent-driven pipelines and DevOps engineers managing agent workflows get direct network access to an agent's capabilities and live task updates.
What does a2acli actually connect and control?
a2acli queries AgentCards to discover an A2A agent's declared skills and transports, then creates and monitors task lifecycles over supported transports. The tool supports REST and WebSocket transports and returns deterministic JSON or NDJSON, which lets other tools parse results reliably. The CLI exposes both an interactive terminal UI for exploration and programmatic modes for automated scripts and AI coding assistants to submit work and fetch structured outputs.
How does it behave in automated workflows and machine use?
The CLI is designed for machine consumption, producing deterministic JSON/NDJSON so downstream parsers and AI models can read outputs without ad hoc parsing. It explicitly targets integration with coding assistants and automated shells, meaning the output format and transport options match the expectations of scripted pipelines. A local mock agent server is included to run development tests without contacting remote agents, enabling repeatable integration tests.
Is it safe to interact with remote agents using this tool?
Agent discovery uses AgentCards, which include security requirements and available transports; a2acli fetches and parses these documents before initiating tasks. That design places security metadata in the discovery step, so users see an agent's declared constraints before sending work. The presence of a mock server also lets teams test workflows locally before connecting to remote agents that declare stricter requirements.
Does it require specialist knowledge to operate correctly?
The CLI surface offers both an interactive TUI for manual exploration and machine-oriented output for automation, so casual users can explore agents while engineers script against NDJSON responses. Support for multiple transports means operators must choose the correct transport and parse JSON streams, which implies familiarity with HTTP/WebSocket interactions and line-delimited JSON for production automation.
Practical recommendation for system integrators and agent developers
a2acli is a pragmatic, developer-focused client for interacting with A2A agents: it exposes discovery, task management, and streaming outputs in formats meant for automation. Teams that build or manage agent-driven pipelines gain predictable outputs and a local test server for validation. Recommended.
