agent-can 0.1.0

Agent-first CAN control daemon
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

# agent-can

Agent-first CAN bus session frontend for real devices. Designed to be consumed by LLM agents over MCP or CLI.

## Quick start

```sh
nix develop -c cargo build

agent-can connect \
  --adapter usb1 \
  --bitrate 500000 \
  --dbc vehicle=./vehicle.dbc

agent-can status
agent-can schema --filter 'vehicle.*'
agent-can message list --allow-raw
agent-can message read --select vehicle.Foo
agent-can message send --target 0x123 --data DEADBEEF
agent-can message send --target vehicle.Foo --data '{"enable":1,"torque":12}'
agent-can trace start --path ./session.asc
agent-can disconnect
```

MCP mode: `agent-can --mcp` exposes the same verbs as MCP tools over stdio.

## Architecture

```text
CLI or MCP  →  local JSON-lines IPC  →  session daemon
                                        - CAN backend (PEAK / SocketCAN)
                                        - connect-time DBC alias registry
                                        - raw event buffer + latest-value index
                                        - periodic send schedule table
                                        - raw ASCII trace export
```

One live session per daemon. The daemon owns all hardware and observation state. CLI and MCP are stateless transports over a shared request contract.

## Platform support

| Platform | Backend | Status |
|----------|---------|--------|
| Windows | PEAK CAN (USB/PCI) | Primary target |
| Linux | SocketCAN | Supported |
| macOS | — | Compiles, no hardware backend |