Tutorial — Agents
Build an AI coding assistant
Bind a Tenzro machine identity to a coding model, expose it through MCP and A2A, and register the built-in
code-review skill so other agents can discover and call it.- Level
- Intermediate
- Time
- ~25 min
- Prerequisites
- Machine identity, served model
- Stack
- CLI · Python
01
Serve a coding model
Qwen 3 0.6B handles general coding; switch to DeepSeek V3 or Phi 3 for harder problems.
tenzro model serve qwen-3-0.6b02
Register the code-review skill
The code-review skill ships built-in; bind it to your machine DID so calls land on you.
tenzro skill register \
--name code-review \
--endpoint "tenzro://did:tenzro:machine:.../code-review"03
Expose A2A and MCP surfaces
The node already serves the A2A Agent Card at a2a.tenzro.network/.well-known/agent.json; add the coding skill to its declared capabilities.
tenzro agent update-card \
--add-skill code-review \
--add-skill inference04
Call the assistant from another agent
Any A2A-aware caller can dispatch a coding task by referencing the skill name.
curl -s https://a2a.tenzro.network/a2a \
-H 'content-type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tasks/send","params":{"skill":"code-review","input":{"repo":"...","pr":42}}}'Related