P3: Fix plugin tests, add GitHub Pages site, update A2A docs

This commit is contained in:
John Doe
2026-03-31 13:13:25 -04:00
parent e1297cde55
commit e1010e4a51
5 changed files with 259 additions and 17 deletions
+76 -3
View File
@@ -4,8 +4,51 @@ _Environment-specific configuration for the Alpha agent._
## A2A Communication
- **Gateway:** `http://localhost:4000`
- **Agent Endpoints:** `/v1/agents/{agent_name}/send`
### Gateway WebSocket RPC
- **Gateway Endpoint:** `ws://127.0.0.1:18789`
- **WebSocket Subprotocol:** `a2a-v1`
- **Message Format:** A2A Protocol v1.0.0
### Connection Example
```javascript
const WebSocket = require('ws');
const ws = new WebSocket('ws://127.0.0.1:18789', ['a2a-v1']);
ws.on('open', () => {
// Send handshake
ws.send(JSON.stringify({
type: 'handshake',
content: {
action: 'advertise',
capabilities: {
supportedMessageTypes: ['message', 'status', 'error', 'proposal', 'vote', 'decision'],
version: '1.0.0'
}
}
}));
});
```
### Message Types Used
| Type | Code | Purpose |
|------|------|---------|
| `message` | 0x01 | Send/receive agent messages |
| `status` | 0x02 | Broadcast status updates |
| `proposal` | 0x30 | Receive deliberation proposals |
| `vote` | 0x32 | Submit triad votes |
| `decision` | 0x31 | Receive triad decisions |
### LiteLLM Integration (Model Routing Only)
- **LiteLLM Gateway:** `http://localhost:4000`
- **Agent Passthrough Endpoint:** `/v1/agents/alpha/send`
- **Health Check:** `/health`
**Note:** LiteLLM is used for model routing only, NOT for A2A communication.
## Triad Sessions
@@ -20,6 +63,36 @@ _Environment-specific configuration for the Alpha agent._
- Threshold: 2 of 3 (66%)
- Decision record: consensus ledger
## Triad Deliberation Protocol
### Vote Message Format
```json
{
"type": "vote",
"from": "alpha",
"content": {
"proposalId": "prop-001",
"vote": "approve" | "reject" | "abstain",
"reasoning": "Feature aligns with roadmap"
},
"metadata": {
"correlationId": "prop-001"
}
}
```
### Consensus Rules
| Votes | Result |
|-------|--------|
| 3-0 | Approved |
| 2-1 | Approved |
| 2-0-1 | Approved |
| 1-2 | Rejected |
| 0-3 | Rejected |
| 1-1-1 | Rejected (no consensus) |
---
*Alpha — Triad Node*
*Alpha — Triad Node*
+47 -4
View File
@@ -4,8 +4,51 @@ _Environment-specific configuration for the Explorer agent._
## A2A Communication
- **Gateway:** `http://localhost:4000`
- **Agent Endpoints:** `/v1/agents/{agent_name}/send`
### Gateway WebSocket RPC
- **Gateway Endpoint:** `ws://127.0.0.1:18789`
- **WebSocket Subprotocol:** `a2a-v1`
- **Message Format:** A2A Protocol v1.0.0
### Connection Example
```javascript
const WebSocket = require('ws');
const ws = new WebSocket('ws://127.0.0.1:18789', ['a2a-v1']);
ws.on('open', () => {
// Send handshake
ws.send(JSON.stringify({
type: 'handshake',
content: {
action: 'advertise',
capabilities: {
supportedMessageTypes: ['message', 'status', 'error', 'request', 'response'],
version: '1.0.0'
}
}
}));
});
```
### Message Types Used
| Type | Code | Purpose |
|------|------|---------|
| `message` | 0x01 | Send/receive agent messages |
| `status` | 0x02 | Broadcast status updates |
| `request` | 0x33 | Request data from other agents |
| `response` | 0x34 | Respond to data requests |
| `broadcast` | 0x35 | Share discoveries with collective |
### LiteLLM Integration (Model Routing Only)
- **LiteLLM Gateway:** `http://localhost:4000`
- **Agent Passthrough Endpoint:** `/v1/agents/explorer/send`
- **Health Check:** `/health`
**Note:** LiteLLM is used for model routing only, NOT for A2A communication.
## Monitoring Sources
@@ -17,7 +60,7 @@ _Environment-specific configuration for the Explorer agent._
## Browser Access Integration
The Explorer agent now has browser automation capabilities via the [`browser-access`](../../skills/browser-access/SKILL.md) skill:
The Explorer agent has browser automation capabilities via the [`browser-access`](../../skills/browser-access/SKILL.md) skill:
### Capabilities
@@ -84,4 +127,4 @@ See [`skills/browser-access/.env.example`](../../skills/browser-access/.env.exam
🧭
*Explorer — Scout*
*Explorer — Scout*
+46 -4
View File
@@ -8,14 +8,56 @@ _Environment-specific configuration and available tools for the Historian agent.
name: Historian
role: memory-keeper
specialization: knowledge-archival
port: 8010
model: minimax/abab6.5s-chat
model: minimax/MiniMax-M2.7
```
## A2A Communication
- **Gateway:** `http://localhost:4000`
- **Agent Endpoints:** `/v1/agents/{agent_name}/send`
### Gateway WebSocket RPC
- **Gateway Endpoint:** `ws://127.0.0.1:18789`
- **WebSocket Subprotocol:** `a2a-v1`
- **Message Format:** A2A Protocol v1.0.0
### Connection Example
```javascript
const WebSocket = require('ws');
const ws = new WebSocket('ws://127.0.0.1:18789', ['a2a-v1']);
ws.on('open', () => {
// Send handshake
ws.send(JSON.stringify({
type: 'handshake',
content: {
action: 'advertise',
capabilities: {
supportedMessageTypes: ['message', 'status', 'error', 'request', 'response'],
version: '1.0.0'
}
}
}));
});
```
### Message Types Used
| Type | Code | Purpose |
|------|------|---------|
| `message` | 0x01 | Send/receive agent messages |
| `status` | 0x02 | Broadcast status updates |
| `request` | 0x33 | Query historical data |
| `response` | 0x34 | Return historical results |
| `broadcast` | 0x35 | Share archival updates |
### LiteLLM Integration (Model Routing Only)
- **LiteLLM Gateway:** `http://localhost:4000`
- **Agent Passthrough Endpoint:** `/v1/agents/historian/send`
- **Health Check:** `/health`
**Note:** LiteLLM is used for model routing only, NOT for A2A communication.
### Input Channels
+46 -3
View File
@@ -4,8 +4,51 @@ _Environment-specific configuration for the Sentinel agent._
## A2A Communication
- **Gateway:** `http://localhost:4000`
- **Agent Endpoints:** `/v1/agents/{agent_name}/send`
### Gateway WebSocket RPC
- **Gateway Endpoint:** `ws://127.0.0.1:18789`
- **WebSocket Subprotocol:** `a2a-v1`
- **Message Format:** A2A Protocol v1.0.0
### Connection Example
```javascript
const WebSocket = require('ws');
const ws = new WebSocket('ws://127.0.0.1:18789', ['a2a-v1']);
ws.on('open', () => {
// Send handshake
ws.send(JSON.stringify({
type: 'handshake',
content: {
action: 'advertise',
capabilities: {
supportedMessageTypes: ['message', 'status', 'error', 'proposal', 'vote'],
version: '1.0.0'
}
}
}));
});
```
### Message Types Used
| Type | Code | Purpose |
|------|------|---------|
| `message` | 0x01 | Send/receive agent messages |
| `status` | 0x02 | Broadcast status updates |
| `error` | 0x03 | Report security errors |
| `proposal` | 0x30 | Submit safety proposals |
| `vote` | 0x32 | Vote on triad decisions |
### LiteLLM Integration (Model Routing Only)
- **LiteLLM Gateway:** `http://localhost:4000`
- **Agent Passthrough Endpoint:** `/v1/agents/sentinel/send`
- **Health Check:** `/health`
**Note:** LiteLLM is used for model routing only, NOT for A2A communication.
## Review Focus
@@ -86,4 +129,4 @@ const stats = shield.getStats();
🦔
*Sentinel — Guardian*
*Sentinel — Guardian*
+44 -3
View File
@@ -4,10 +4,51 @@ _Environment-specific configuration for the Steward agent._
## A2A Communication
- **Gateway:** `http://localhost:4000`
- **Agent Endpoints:** `/v1/agents/{agent_name}/send`
### Gateway WebSocket RPC
- **Gateway Endpoint:** `ws://127.0.0.1:18789`
- **WebSocket Subprotocol:** `a2a-v1`
- **Message Format:** A2A Protocol v1.0.0
### Connection Example
```javascript
const WebSocket = require('ws');
const ws = new WebSocket('ws://127.0.0.1:18789', ['a2a-v1']);
ws.on('open', () => {
// Send handshake
ws.send(JSON.stringify({
type: 'handshake',
content: {
action: 'advertise',
capabilities: {
supportedMessageTypes: ['message', 'status', 'error', 'proposal', 'broadcast'],
version: '1.0.0'
}
}
}));
});
```
### Message Types Used
| Type | Code | Purpose |
|------|------|---------|
| `message` | 0x01 | Send/receive agent messages |
| `status` | 0x02 | Broadcast status updates |
| `proposal` | 0x30 | Initiate triad deliberation |
| `broadcast` | 0x35 | Multi-agent announcements |
### LiteLLM Integration (Model Routing Only)
- **LiteLLM Gateway:** `http://localhost:4000`
- **Agent Passthrough Endpoint:** `/v1/agents/steward/send`
- **Health Check:** `/health`
**Note:** LiteLLM is used for model routing only, NOT for A2A communication.
## Agent Sessions
| Agent | Session ID |
@@ -35,4 +76,4 @@ _Environment-specific configuration for the Steward agent._
🦞
*Steward — Orchestrator*
*Steward — Orchestrator*