Conferences · Community · Tools

Agent Engineering HQ¶

The home of the agent engineering discipline.

Meet the people, explore the tools and find the events shaping agent engineering. Take AgentEng with you, from your terminal to the coding agent you already use.

📍 London + San Francisco
🧠 12 disciplines
💻 CLI · MCP · A2A
🔓 Open source

🚀 Get started 🤖 Connect your agent 🌍 Event website

🧭 Find your starting point¶

📅 Discover events

Find London and San Francisco events, speakers, sessions and official registration links.

Explore events →
🤖 Bring your coding agent

Connect through one MCP tool, A2A or HTTP. Use the same catalogue from your existing workspace.

Connect an agent →
🧰 Explore the tool directory

Browse 456 attributed tools, models and infrastructure listings across twelve disciplines.

Browse the directory →
💡 Shape a future event

Turn a talk, workshop or event idea into a local draft, then choose how to share it with the organizer.

Prepare an idea →

🚀 Installation and first run¶

Start with Python 3.12+ and uv. Install from the source checkout, including the optional local MCP server:

Install AgentEng

git clone https://github.com/SuperagenticAI/agenteng.git
cd agenteng
uv tool install '.[mcp]'
agenteng discover

Your first result: published London and San Francisco events, with supporting source links. The CLI uses its bundled snapshot, so event lookup, tool browsing and local drafting work offline with zero model calls and no provider key.

The package and command are both named agenteng. This is a 0.1.0 alpha. Run agenteng --help to explore commands, or agenteng COMMAND --help for options. Installing the CLI does not start or publish a hosted service.

📅 Explore events¶

Find an event, then use its returned ID to explore the agenda, speakers or ticket terms. Put global options such as --json before the command.

Find your next event

agenteng events --city London --upcoming
agenteng events --city 'San Francisco'
agenteng ask 'When is the next London conference?'
agenteng --json events

Explore a published programme

agenteng speakers --city London
agenteng agenda agenteng-london-2026 --topic memory
agenteng tickets agenteng-london-2026

The catalogue is a dated snapshot. Follow the event's official registration link for current availability and details. See data and attribution.

🧠 Twelve disciplines, one directory¶

Explore the tooling around the whole discipline, from prompts and memory to protocols and inference. Listings include names, categories and source links; they are not popularity rankings or tutorials.

✍️ Prompt
🧩 Context
⚙️ Harness
📏 Eval
🧠 Memory
⚡ Inference
🔄 Loop
🤖 Agentic
💻 Code
🔌 Protocol
🕸️ Graph
🔎 Search

Find tools for your work

agenteng disciplines
agenteng tools --discipline memory
agenteng tools --discipline inference --kind runtime
agenteng tools --search 'Gemini CLI'
agenteng tool langgraph

Explore filters, pagination and source attribution →

🤖 Use AgentEng inside your agent¶

Add the installed executable to a client's local MCP configuration:

Local MCP configuration

{
  "mcpServers": {
    "agenteng": {
      "command": "agenteng",
      "args": ["mcp"]
    }
  }
}

Try asking: “Find the next London conference”, “List memory tools” or “Help me draft a workshop idea for San Francisco.”

Run agenteng connect codex, agenteng connect claude-code or agenteng connect cursor for client-specific setup instructions. The connection guide also covers hosted MCP, A2A and HTTP.

A practical workshop, a talk you want to give, a topic the community should explore: start small and shape the idea before sharing it.

Prepare a future-event idea

agenteng engage --city London --output draft.json
agenteng proposal preview draft.json
agenteng proposal export draft.json --format markdown --output draft.md
agenteng participate

These commands create local drafts and show organizer contact details; they do not send a proposal. Ideas go to one organizing group: Agent Engineering HQ. Sharing an idea does not guarantee a response, acceptance or an event. London 2026 has an invited programme and no public CFP.

Drafting and the optional intake pilot → Community participation →

🔓 Built in the open¶

Improve the code, suggest a source-backed tool correction or help make the docs clearer. AgentEng shares one typed interface across CLI, MCP and A2A. Optional model synthesis and RLM are disabled by default; RLM is bounded to depth one and one child or leaf delegation.