mulesoft-anypoint-cli-agent-fabric-plugin 1.4.0 → 1.5.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/GETTING_STARTED.md +396 -0
- package/README.md +26 -1
- package/TROUBLESHOOTING.md +412 -0
- package/dist/commands/agent-network/project/create.d.ts +6 -2
- package/dist/commands/agent-network/project/create.js +44 -13
- package/dist/commands/agent-network/project/create.js.map +1 -1
- package/dist/commands/agent-network/project/deploy.js +18 -20
- package/dist/commands/agent-network/project/deploy.js.map +1 -1
- package/dist/commands/agent-network/project/publish.js +2 -1
- package/dist/commands/agent-network/project/publish.js.map +1 -1
- package/dist/commands/agent-network/project/run-local.d.ts +10 -0
- package/dist/commands/agent-network/project/run-local.js +168 -47
- package/dist/commands/agent-network/project/run-local.js.map +1 -1
- package/dist/commands/agent-network/project/undeploy.js +16 -9
- package/dist/commands/agent-network/project/undeploy.js.map +1 -1
- package/dist/commands/agent-network/project/unpublish.js +23 -16
- package/dist/commands/agent-network/project/unpublish.js.map +1 -1
- package/dist/errors/base-error.d.ts +1 -0
- package/dist/errors/base-error.js +28 -18
- package/dist/errors/base-error.js.map +1 -1
- package/dist/errors/error-factory.d.ts +36 -3
- package/dist/errors/error-factory.js +184 -8
- package/dist/errors/error-factory.js.map +1 -1
- package/dist/helpers/connection.d.ts +6 -2
- package/dist/helpers/connection.js +13 -6
- package/dist/helpers/connection.js.map +1 -1
- package/dist/helpers/project/constants.d.ts +1 -1
- package/dist/helpers/proxy-agent.d.ts +2 -0
- package/dist/helpers/proxy-agent.js +26 -0
- package/dist/helpers/proxy-agent.js.map +1 -0
- package/dist/helpers/request-logging.d.ts +4 -0
- package/dist/helpers/request-logging.js +96 -2
- package/dist/helpers/request-logging.js.map +1 -1
- package/dist/resources/schemas/broker-deployment-settings.schema.json +4 -0
- package/dist/types/agent-network-v2.d.ts +36 -9
- package/dist/types/agent-network-v2.js.map +1 -1
- package/dist/types/agent-network.d.ts +0 -4
- package/dist/types/agent-network.js.map +1 -1
- package/dist/types/types.d.ts +20 -3
- package/dist/utils/builders/builder.d.ts +2 -0
- package/dist/utils/builders/builder.js +22 -1
- package/dist/utils/builders/builder.js.map +1 -1
- package/dist/utils/builders/common/connection-facet-helpers.d.ts +2 -0
- package/dist/utils/builders/common/connection-facet-helpers.js +4 -1
- package/dist/utils/builders/common/connection-facet-helpers.js.map +1 -1
- package/dist/utils/builders/common/network-builder-helpers.d.ts +5 -0
- package/dist/utils/builders/common/network-builder-helpers.js +8 -0
- package/dist/utils/builders/common/network-builder-helpers.js.map +1 -1
- package/dist/utils/builders/common/network-metadata.js +1 -1
- package/dist/utils/builders/common/network-metadata.js.map +1 -1
- package/dist/utils/builders/common/types.d.ts +1 -1
- package/dist/utils/builders/local-space-resolver.d.ts +13 -3
- package/dist/utils/builders/local-space-resolver.js +16 -3
- package/dist/utils/builders/local-space-resolver.js.map +1 -1
- package/dist/utils/builders/policies-factory.d.ts +5 -1
- package/dist/utils/builders/policies-factory.js +74 -7
- package/dist/utils/builders/policies-factory.js.map +1 -1
- package/dist/utils/builders/project-builder-impl.d.ts +8 -1
- package/dist/utils/builders/project-builder-impl.js +17 -1
- package/dist/utils/builders/project-builder-impl.js.map +1 -1
- package/dist/utils/builders/project-builder.d.ts +8 -1
- package/dist/utils/builders/project-builder.js +2 -1
- package/dist/utils/builders/project-builder.js.map +1 -1
- package/dist/utils/builders/remote-space-resolver.d.ts +3 -3
- package/dist/utils/builders/remote-space-resolver.js +47 -32
- package/dist/utils/builders/remote-space-resolver.js.map +1 -1
- package/dist/utils/builders/v1/broker-builder.js +2 -1
- package/dist/utils/builders/v1/broker-builder.js.map +1 -1
- package/dist/utils/builders/v1/connections-builder.js +3 -0
- package/dist/utils/builders/v1/connections-builder.js.map +1 -1
- package/dist/utils/builders/v1/network-builder.js +10 -2
- package/dist/utils/builders/v1/network-builder.js.map +1 -1
- package/dist/utils/builders/v2/broker-builder.js +1 -1
- package/dist/utils/builders/v2/broker-builder.js.map +1 -1
- package/dist/utils/builders/v2/connections-builder.d.ts +5 -1
- package/dist/utils/builders/v2/connections-builder.js +93 -19
- package/dist/utils/builders/v2/connections-builder.js.map +1 -1
- package/dist/utils/builders/v2/llm-builder.js +17 -6
- package/dist/utils/builders/v2/llm-builder.js.map +1 -1
- package/dist/utils/builders/v2/network-builder.js +2 -0
- package/dist/utils/builders/v2/network-builder.js.map +1 -1
- package/dist/utils/constants.d.ts +15 -1
- package/dist/utils/constants.js +19 -4
- package/dist/utils/constants.js.map +1 -1
- package/dist/utils/deployment-settings.d.ts +1 -0
- package/dist/utils/deployment-settings.js.map +1 -1
- package/dist/utils/dry-run/dry-run-render.d.ts +1 -0
- package/dist/utils/dry-run/dry-run-render.js +23 -13
- package/dist/utils/dry-run/dry-run-render.js.map +1 -1
- package/dist/utils/exchange.d.ts +1 -0
- package/dist/utils/exchange.js +20 -0
- package/dist/utils/exchange.js.map +1 -1
- package/dist/utils/facets/agent-graph-facet.d.ts +1 -1
- package/dist/utils/facets/agent-graph-facet.js +8 -8
- package/dist/utils/facets/agent-graph-facet.js.map +1 -1
- package/dist/utils/facets/api-instance-facet.d.ts +2 -0
- package/dist/utils/facets/api-instance-facet.js +72 -14
- package/dist/utils/facets/api-instance-facet.js.map +1 -1
- package/dist/utils/facets/apim-policy-helpers.d.ts +29 -2
- package/dist/utils/facets/apim-policy-helpers.js +19 -5
- package/dist/utils/facets/apim-policy-helpers.js.map +1 -1
- package/dist/utils/facets/asset-facet.d.ts +1 -0
- package/dist/utils/facets/asset-facet.js +34 -25
- package/dist/utils/facets/asset-facet.js.map +1 -1
- package/dist/utils/facets/connection-facet.d.ts +6 -0
- package/dist/utils/facets/connection-facet.js +199 -17
- package/dist/utils/facets/connection-facet.js.map +1 -1
- package/dist/utils/facets/deployment-helpers.d.ts +1 -0
- package/dist/utils/facets/deployment-helpers.js +26 -20
- package/dist/utils/facets/deployment-helpers.js.map +1 -1
- package/dist/utils/facets/deployment-runner.js +4 -1
- package/dist/utils/facets/deployment-runner.js.map +1 -1
- package/dist/utils/facets/facets.d.ts +1 -0
- package/dist/utils/facets/facets.js +16 -7
- package/dist/utils/facets/facets.js.map +1 -1
- package/dist/utils/facets/mule-application-facet.js +2 -2
- package/dist/utils/facets/mule-application-facet.js.map +1 -1
- package/dist/utils/facets/secret-manager-client.d.ts +16 -0
- package/dist/utils/facets/secret-manager-client.js +67 -0
- package/dist/utils/facets/secret-manager-client.js.map +1 -0
- package/dist/utils/facets/secret-tls-validator.d.ts +5 -0
- package/dist/utils/facets/secret-tls-validator.js +60 -0
- package/dist/utils/facets/secret-tls-validator.js.map +1 -0
- package/dist/utils/gateway.js +1 -1
- package/dist/utils/gateway.js.map +1 -1
- package/dist/utils/run-local/build-status.js +2 -2
- package/dist/utils/run-local/build-status.js.map +1 -1
- package/dist/utils/run-local/index.d.ts +1 -1
- package/dist/utils/run-local/index.js +2 -1
- package/dist/utils/run-local/index.js.map +1 -1
- package/dist/utils/run-local/local-a2a-upstream.d.ts +1 -0
- package/dist/utils/run-local/local-a2a-upstream.js +5 -4
- package/dist/utils/run-local/local-a2a-upstream.js.map +1 -1
- package/dist/utils/run-local/local-tls-model.d.ts +18 -0
- package/dist/utils/run-local/local-tls-model.js +97 -0
- package/dist/utils/run-local/local-tls-model.js.map +1 -0
- package/dist/utils/run-local/local-tls.d.ts +37 -0
- package/dist/utils/run-local/local-tls.js +245 -0
- package/dist/utils/run-local/local-tls.js.map +1 -0
- package/dist/utils/run-local/policy-resolver.js +5 -0
- package/dist/utils/run-local/policy-resolver.js.map +1 -1
- package/dist/utils/run-local/runtime-environment-manager.d.ts +2 -2
- package/dist/utils/run-local/runtime-environment-manager.js +8 -4
- package/dist/utils/run-local/runtime-environment-manager.js.map +1 -1
- package/dist/utils/run-local/secrets-manager.d.ts +1 -0
- package/dist/utils/run-local/secrets-manager.js +2 -1
- package/dist/utils/run-local/secrets-manager.js.map +1 -1
- package/dist/utils/sanitize.d.ts +1 -0
- package/dist/utils/sanitize.js +14 -0
- package/dist/utils/sanitize.js.map +1 -0
- package/dist/utils/variables.d.ts +7 -2
- package/dist/utils/variables.js +46 -6
- package/dist/utils/variables.js.map +1 -1
- package/package.json +6 -4
- package/scripts/extract-error-catalog.js +88 -0
- package/templates/agentic-network/agent-network.yaml.template +1 -7
- package/templates/agentic-network/brokers/broker1.agent.template +1 -1
- package/templates/agentic-network/exchange.json.template +9 -1
|
@@ -0,0 +1,396 @@
|
|
|
1
|
+
# Getting started with an Agent Network project
|
|
2
|
+
|
|
3
|
+
This guide walks through the files `agent-network project create` scaffolds, so you know what
|
|
4
|
+
each section is for before you start editing. It matches the default create output — create a
|
|
5
|
+
project yourself and follow along:
|
|
6
|
+
|
|
7
|
+
```bash
|
|
8
|
+
anypoint-cli-agent-fabric-plugin agent-network project create --name my-agent-network
|
|
9
|
+
```
|
|
10
|
+
|
|
11
|
+
If you hit an error along the way, see [TROUBLESHOOTING.md](TROUBLESHOOTING.md).
|
|
12
|
+
|
|
13
|
+
## Project layout
|
|
14
|
+
|
|
15
|
+
```
|
|
16
|
+
my-agent-network/
|
|
17
|
+
├── exchange.json # Exchange descriptor: GAV, classifier, variables
|
|
18
|
+
├── agent-network.yaml # topology: registry, connections, brokers
|
|
19
|
+
└── brokers/
|
|
20
|
+
└── broker1.agent # AgentScript implementation for broker1
|
|
21
|
+
```
|
|
22
|
+
|
|
23
|
+
## `exchange.json` — the Exchange descriptor
|
|
24
|
+
|
|
25
|
+
Every Agent Network project is published to Exchange as an asset, so it needs a descriptor:
|
|
26
|
+
|
|
27
|
+
```json
|
|
28
|
+
{
|
|
29
|
+
"name": "my-agent-network",
|
|
30
|
+
"classifier": "agentic-network",
|
|
31
|
+
"groupId": "<your-org-id>",
|
|
32
|
+
"assetId": "my-agent-network",
|
|
33
|
+
"version": "1.0.0",
|
|
34
|
+
"dependencies": [],
|
|
35
|
+
"metadata": {
|
|
36
|
+
"variables": {
|
|
37
|
+
"myAgent": {
|
|
38
|
+
"url": { "description": "My External Agent URL", "default": "", "secret": false }
|
|
39
|
+
},
|
|
40
|
+
"openai": {
|
|
41
|
+
"apiKey": { "description": "OpenAI API key", "default": "" }
|
|
42
|
+
}
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
}
|
|
46
|
+
```
|
|
47
|
+
|
|
48
|
+
- **`groupId` / `assetId` / `version`** form the project's GAV — its coordinate in Exchange.
|
|
49
|
+
- **`classifier: "agentic-network"`** marks this as an Agent Network asset (as opposed to, say, a
|
|
50
|
+
connector or a Mule application).
|
|
51
|
+
- **`dependencies`** lists Exchange assets the project consumes. The default create scaffold leaves
|
|
52
|
+
this empty and declares OpenAI inline under `registry.llms` (see below). You can instead add an
|
|
53
|
+
`llm-metadata` dependency (for example the published `openai` asset) and point connections at it
|
|
54
|
+
with a namespaced `ref`.
|
|
55
|
+
- **`metadata.variables`** declares placeholders that `agent-network.yaml` can reference as
|
|
56
|
+
`${namespace.name}` (for example `${myAgent.url}`, `${openai.apiKey}`). Values are supplied at
|
|
57
|
+
deploy time via `--property namespace.name:value` or an interactive prompt — nothing sensitive
|
|
58
|
+
needs to live in the project files. Mark a variable `"secret": true` if its value shouldn't be
|
|
59
|
+
echoed back or logged.
|
|
60
|
+
|
|
61
|
+
## `agent-network.yaml` — the topology
|
|
62
|
+
|
|
63
|
+
This is the file that describes what the network is made of and how the pieces connect.
|
|
64
|
+
|
|
65
|
+
### `registry` — the building blocks
|
|
66
|
+
|
|
67
|
+
```yaml
|
|
68
|
+
registry:
|
|
69
|
+
agents:
|
|
70
|
+
myAgent:
|
|
71
|
+
info:
|
|
72
|
+
label: My External Agent
|
|
73
|
+
metadata:
|
|
74
|
+
platform: Bedrock
|
|
75
|
+
interfaces:
|
|
76
|
+
a2a:
|
|
77
|
+
card: { ... } # an A2A AgentCard describing capabilities and skills
|
|
78
|
+
mcps:
|
|
79
|
+
myMcpServer:
|
|
80
|
+
info:
|
|
81
|
+
label: My MCP Server
|
|
82
|
+
metadata:
|
|
83
|
+
transport:
|
|
84
|
+
kind: streamableHttp
|
|
85
|
+
path: "/mcp"
|
|
86
|
+
llms:
|
|
87
|
+
openai:
|
|
88
|
+
info:
|
|
89
|
+
label: OpenAI
|
|
90
|
+
description: OpenAI LLM Provider
|
|
91
|
+
metadata:
|
|
92
|
+
platform: OpenAI
|
|
93
|
+
```
|
|
94
|
+
|
|
95
|
+
- **`registry.agents`** — external agents this network can call. Each entry's `metadata.interfaces`
|
|
96
|
+
describes the protocol used to reach it (A2A is the one shown here).
|
|
97
|
+
- **`registry.mcps`** — Model Context Protocol servers that expose tools, and how to reach them
|
|
98
|
+
(transport kind and path).
|
|
99
|
+
- **`registry.llms`** — LLM providers declared inline in the project. The default create scaffold
|
|
100
|
+
uses this for OpenAI. Alternatively, omit `registry.llms` and consume a published
|
|
101
|
+
`llm-metadata` asset from `exchange.json` `dependencies`, referencing it from the connection
|
|
102
|
+
with `ref.namespace` set to that asset's group id.
|
|
103
|
+
|
|
104
|
+
Every entry here is a *declaration*; how the network actually reaches each one at runtime is
|
|
105
|
+
defined next, under `context.connections`.
|
|
106
|
+
|
|
107
|
+
### `context.connections` — how to reach each registry entry
|
|
108
|
+
|
|
109
|
+
```yaml
|
|
110
|
+
context:
|
|
111
|
+
connections:
|
|
112
|
+
myAgentConnection:
|
|
113
|
+
kind: a2a
|
|
114
|
+
ref: { name: myAgent }
|
|
115
|
+
url: ${myAgent.url}
|
|
116
|
+
myMcpServerConnection:
|
|
117
|
+
kind: mcp
|
|
118
|
+
ref: { name: myMcpServer }
|
|
119
|
+
url: ${myMcpServer.url}
|
|
120
|
+
openaiConnection:
|
|
121
|
+
kind: llm
|
|
122
|
+
ref: { name: openai }
|
|
123
|
+
url: https://api.openai.com/v1
|
|
124
|
+
authentication:
|
|
125
|
+
kind: apiKey
|
|
126
|
+
apiKey: ${openai.apiKey}
|
|
127
|
+
```
|
|
128
|
+
|
|
129
|
+
Each connection names a `kind` (`a2a`, `mcp`, or `llm`), a `ref` back to the registry entry (or
|
|
130
|
+
Exchange asset) it serves, a `url` (often a `${variable}` resolved at deploy time), and — for
|
|
131
|
+
LLMs and any authenticated target — an `authentication` block. For an inline `registry.llms`
|
|
132
|
+
entry, `ref` is just `{ name: openai }`. For an Exchange `llm-metadata` dependency, add
|
|
133
|
+
`namespace: <group-id>` on the `ref`. Brokers reference connections by name (e.g.
|
|
134
|
+
`a2a://myAgentConnection`) rather than embedding URLs directly, so credentials and endpoints stay
|
|
135
|
+
out of the broker logic.
|
|
136
|
+
|
|
137
|
+
### `brokers` — the orchestrating agents
|
|
138
|
+
|
|
139
|
+
```yaml
|
|
140
|
+
brokers:
|
|
141
|
+
broker1:
|
|
142
|
+
kind: AgentScript
|
|
143
|
+
implementation: ./brokers/broker1.agent
|
|
144
|
+
interfaces:
|
|
145
|
+
a2a:
|
|
146
|
+
card:
|
|
147
|
+
name: Broker 1
|
|
148
|
+
description: Describe what your broker does.
|
|
149
|
+
skills:
|
|
150
|
+
- id: my-skill
|
|
151
|
+
name: My Skill
|
|
152
|
+
```
|
|
153
|
+
|
|
154
|
+
A broker is exposed as its own A2A agent (via `interfaces.a2a.card`) and its behavior lives in a
|
|
155
|
+
separate `AgentScript` file referenced by `implementation`.
|
|
156
|
+
|
|
157
|
+
## The AgentScript broker file
|
|
158
|
+
|
|
159
|
+
`brokers/broker1.agent` is where the orchestration logic lives. A minimal broker has these
|
|
160
|
+
sections, roughly in the order execution flows through them:
|
|
161
|
+
|
|
162
|
+
| Section | Purpose |
|
|
163
|
+
| --- | --- |
|
|
164
|
+
| `system` | Top-level instructions/persona for the broker. |
|
|
165
|
+
| `config` | Broker-level settings, e.g. which `llm` entry is the default. |
|
|
166
|
+
| `llm` | Declares LLM targets, pointing at a `context.connections` entry (`target: "llm://openaiConnection"`). |
|
|
167
|
+
| `actions` | Declares callable actions — an A2A call to another agent (`target: "a2a://..."`) or an MCP tool invocation (`target: "mcp://..."`). |
|
|
168
|
+
| `trigger <name>` | Entry point — how the broker receives work (e.g. an incoming A2A message) and which `orchestrator` node it transitions to. |
|
|
169
|
+
| `orchestrator <name>` | A reasoning step: given the incoming request, decide which `actions` to invoke and what to output. |
|
|
170
|
+
| `generator <name>` | Formats a response (e.g. turning orchestrator output into user-facing text) using an LLM call. |
|
|
171
|
+
| `executor <name>` | Actually runs an action (e.g. calls an MCP tool) with resolved inputs. |
|
|
172
|
+
| `router <name>` | Branches to different `echo` nodes based on a condition (e.g. success vs. failure). |
|
|
173
|
+
| `echo <name>` | Terminal node — emits the final A2A response back to the caller. |
|
|
174
|
+
|
|
175
|
+
Nodes transition to each other via `on_exit: -> transition to @<section>.<name>`, and reference
|
|
176
|
+
each other's outputs via `@<section>.<name>.output`. Reading a broker file top-to-bottom in
|
|
177
|
+
declaration order won't reflect execution order — follow the `transition to` chain starting from
|
|
178
|
+
the `trigger` block instead.
|
|
179
|
+
|
|
180
|
+
## Full worked example: a weather-lookup broker
|
|
181
|
+
|
|
182
|
+
The snippets above show each piece in isolation. Here's a complete, self-contained network that
|
|
183
|
+
ties them together: `weatherBroker` takes a natural-language question, calls a weather MCP tool
|
|
184
|
+
to fetch the data, then uses an LLM to summarize it back to the user. Every field below is filled
|
|
185
|
+
in — the only values you need to supply are the two declared variables (`openai.apiKey` and
|
|
186
|
+
`weatherMcp.url`, via `--property` at deploy time).
|
|
187
|
+
|
|
188
|
+
### `exchange.json`
|
|
189
|
+
|
|
190
|
+
```json
|
|
191
|
+
{
|
|
192
|
+
"main": "agent-network.yaml",
|
|
193
|
+
"name": "weather-assistant",
|
|
194
|
+
"classifier": "agentic-network",
|
|
195
|
+
"organizationId": "68ef9520-24e9-4098-93bb-081c75e21a26",
|
|
196
|
+
"descriptorVersion": "1.0.0",
|
|
197
|
+
"apiVersion": "1.0.0",
|
|
198
|
+
"tags": [],
|
|
199
|
+
"groupId": "68ef9520-24e9-4098-93bb-081c75e21a26",
|
|
200
|
+
"assetId": "weather-assistant",
|
|
201
|
+
"version": "1.0.0",
|
|
202
|
+
"dependencies": [],
|
|
203
|
+
"metadata": {
|
|
204
|
+
"toolingVersion": 1,
|
|
205
|
+
"variables": {
|
|
206
|
+
"openai": {
|
|
207
|
+
"apiKey": { "description": "OpenAI API key", "default": "", "secret": true }
|
|
208
|
+
},
|
|
209
|
+
"weatherMcp": {
|
|
210
|
+
"url": { "description": "Weather MCP server URL", "default": "", "secret": false }
|
|
211
|
+
}
|
|
212
|
+
}
|
|
213
|
+
}
|
|
214
|
+
}
|
|
215
|
+
```
|
|
216
|
+
|
|
217
|
+
### `agent-network.yaml`
|
|
218
|
+
|
|
219
|
+
```yaml
|
|
220
|
+
agentNetwork: 2.0.0
|
|
221
|
+
info:
|
|
222
|
+
label: Weather Assistant
|
|
223
|
+
version: v1
|
|
224
|
+
registry:
|
|
225
|
+
mcps:
|
|
226
|
+
weatherMcp:
|
|
227
|
+
info:
|
|
228
|
+
label: Weather MCP Server
|
|
229
|
+
metadata:
|
|
230
|
+
transport:
|
|
231
|
+
kind: streamableHttp
|
|
232
|
+
path: "/mcp"
|
|
233
|
+
llms:
|
|
234
|
+
openai:
|
|
235
|
+
info:
|
|
236
|
+
label: OpenAI
|
|
237
|
+
description: OpenAI LLM Provider
|
|
238
|
+
metadata:
|
|
239
|
+
platform: OpenAI
|
|
240
|
+
context:
|
|
241
|
+
connections:
|
|
242
|
+
weatherMcpConnection:
|
|
243
|
+
kind: mcp
|
|
244
|
+
ref:
|
|
245
|
+
name: weatherMcp
|
|
246
|
+
url: ${weatherMcp.url}
|
|
247
|
+
openaiConnection:
|
|
248
|
+
kind: llm
|
|
249
|
+
ref:
|
|
250
|
+
name: openai
|
|
251
|
+
url: https://api.openai.com/v1
|
|
252
|
+
authentication:
|
|
253
|
+
kind: apiKey
|
|
254
|
+
apiKey: ${openai.apiKey}
|
|
255
|
+
brokers:
|
|
256
|
+
weatherBroker:
|
|
257
|
+
kind: AgentScript
|
|
258
|
+
implementation: ./brokers/weatherBroker.agent
|
|
259
|
+
interfaces:
|
|
260
|
+
a2a:
|
|
261
|
+
card:
|
|
262
|
+
name: Weather Assistant
|
|
263
|
+
description: Answers questions about current weather by calling a weather MCP tool.
|
|
264
|
+
version: 1.0.0
|
|
265
|
+
capabilities:
|
|
266
|
+
streaming: true
|
|
267
|
+
pushNotifications: false
|
|
268
|
+
defaultInputModes:
|
|
269
|
+
- text/plain
|
|
270
|
+
defaultOutputModes:
|
|
271
|
+
- text/plain
|
|
272
|
+
skills:
|
|
273
|
+
- id: get-weather
|
|
274
|
+
name: Get Weather
|
|
275
|
+
description: Look up the current weather for a city.
|
|
276
|
+
tags:
|
|
277
|
+
- weather
|
|
278
|
+
```
|
|
279
|
+
|
|
280
|
+
### `brokers/weatherBroker.agent`
|
|
281
|
+
|
|
282
|
+
```
|
|
283
|
+
# @dialect: AGENTFABRIC=1.1
|
|
284
|
+
system:
|
|
285
|
+
instructions: "You are a helpful weather assistant."
|
|
286
|
+
|
|
287
|
+
config:
|
|
288
|
+
agent_name: "weatherBroker"
|
|
289
|
+
default_llm: @llm.openai_llm
|
|
290
|
+
|
|
291
|
+
llm:
|
|
292
|
+
openai_llm:
|
|
293
|
+
target: "llm://openaiConnection"
|
|
294
|
+
kind: "OpenAI"
|
|
295
|
+
model: "gpt-4o-mini"
|
|
296
|
+
|
|
297
|
+
actions:
|
|
298
|
+
get_weather:
|
|
299
|
+
target: "mcp://weatherMcpConnection"
|
|
300
|
+
kind: "mcp:tool"
|
|
301
|
+
tool_name: "get-weather"
|
|
302
|
+
inputs:
|
|
303
|
+
city: string
|
|
304
|
+
|
|
305
|
+
trigger weatherTrigger:
|
|
306
|
+
kind: "a2a"
|
|
307
|
+
target: "brokers://weatherBroker/a2a"
|
|
308
|
+
on_message: ->
|
|
309
|
+
transition to @orchestrator.main
|
|
310
|
+
|
|
311
|
+
orchestrator main:
|
|
312
|
+
description: "Decide which city to look up and call the weather tool"
|
|
313
|
+
label: "Main"
|
|
314
|
+
system:
|
|
315
|
+
instructions: |
|
|
316
|
+
Extract the city name from the user's message. Always call
|
|
317
|
+
get_weather when a city is present.
|
|
318
|
+
reasoning:
|
|
319
|
+
instructions: ->
|
|
320
|
+
| {!@request.payload.message.parts[0].text}
|
|
321
|
+
actions:
|
|
322
|
+
get_weather: @actions.get_weather
|
|
323
|
+
with city = ...
|
|
324
|
+
outputs:
|
|
325
|
+
properties:
|
|
326
|
+
city:
|
|
327
|
+
type: "string"
|
|
328
|
+
description: "The city that was looked up"
|
|
329
|
+
status:
|
|
330
|
+
type: "string"
|
|
331
|
+
description: "success or failure"
|
|
332
|
+
on_exit: ->
|
|
333
|
+
transition to @executor.runWeatherLookup
|
|
334
|
+
|
|
335
|
+
executor runWeatherLookup:
|
|
336
|
+
do: ->
|
|
337
|
+
run @actions.get_weather
|
|
338
|
+
with city = @orchestrator.main.output.city
|
|
339
|
+
on_exit: ->
|
|
340
|
+
transition to @generator.formatResponse
|
|
341
|
+
|
|
342
|
+
generator formatResponse:
|
|
343
|
+
system:
|
|
344
|
+
instructions: "You turn raw weather data into a short, friendly summary."
|
|
345
|
+
prompt: ->
|
|
346
|
+
| Summarize this weather data for the user in one or two sentences:
|
|
347
|
+
{!@executor.runWeatherLookup.output}
|
|
348
|
+
on_exit: ->
|
|
349
|
+
transition to @router.resultRouter
|
|
350
|
+
|
|
351
|
+
router resultRouter:
|
|
352
|
+
routes:
|
|
353
|
+
- target: @echo.successResponse
|
|
354
|
+
when: @orchestrator.main.output.status == "success"
|
|
355
|
+
label: "Success"
|
|
356
|
+
otherwise:
|
|
357
|
+
target: @echo.errorResponse
|
|
358
|
+
|
|
359
|
+
echo successResponse:
|
|
360
|
+
kind: "a2a:status_update_event"
|
|
361
|
+
state: "TASK_STATE_COMPLETED"
|
|
362
|
+
message: a2a.message({
|
|
363
|
+
messageId: uuid(),
|
|
364
|
+
parts:[
|
|
365
|
+
a2a.textPart(@generator.formatResponse.output)
|
|
366
|
+
]
|
|
367
|
+
})
|
|
368
|
+
|
|
369
|
+
echo errorResponse:
|
|
370
|
+
kind: "a2a:status_update_event"
|
|
371
|
+
state: "TASK_STATE_FAILED"
|
|
372
|
+
message: a2a.message({
|
|
373
|
+
messageId: uuid(),
|
|
374
|
+
parts:[
|
|
375
|
+
a2a.textPart("I couldn't find weather for that city. Please try again.")
|
|
376
|
+
]
|
|
377
|
+
})
|
|
378
|
+
```
|
|
379
|
+
|
|
380
|
+
Drop these three files into a project (`exchange.json` and `agent-network.yaml` at the root,
|
|
381
|
+
the broker file under `brokers/`), supply real values for `openai.apiKey` and `weatherMcp.url`
|
|
382
|
+
against a running weather MCP server, and `build` / `publish` / `deploy` as usual.
|
|
383
|
+
|
|
384
|
+
## Next steps
|
|
385
|
+
|
|
386
|
+
Once the topology and broker logic look right:
|
|
387
|
+
|
|
388
|
+
```bash
|
|
389
|
+
anypoint-cli-agent-fabric-plugin agent-network project build
|
|
390
|
+
anypoint-cli-agent-fabric-plugin agent-network project publish
|
|
391
|
+
anypoint-cli-agent-fabric-plugin agent-network project deploy
|
|
392
|
+
```
|
|
393
|
+
|
|
394
|
+
Each stage validates and persists its output before the next one runs — see the main
|
|
395
|
+
[README](README.md#usage) for the full command reference, and
|
|
396
|
+
[TROUBLESHOOTING.md](TROUBLESHOOTING.md) if a command fails.
|
package/README.md
CHANGED
|
@@ -145,7 +145,7 @@ FLAGS
|
|
|
145
145
|
--[no-]collectMetrics collect metrics. You can define this in the COLLECT_METRICS environment
|
|
146
146
|
variable.
|
|
147
147
|
--deployment-properties-file=<value> Path to a YAML file of typed deployment settings, keyed by scope (e.g. a
|
|
148
|
-
top-level "brokers:" mapping vCores, resources.*, logging.level,
|
|
148
|
+
top-level "brokers:" mapping vCores, replicas, resources.*, logging.level,
|
|
149
149
|
runtimeVersion). Mutually exclusive with --deployment-property.
|
|
150
150
|
--deployment-property=<value>... [default: ] Typed deployment setting in format <scope>:<name>:<value> (e.g.,
|
|
151
151
|
--deployment-property brokers:vCores:0.1). Repeatable. Mutually exclusive with
|
|
@@ -180,8 +180,25 @@ EXAMPLES
|
|
|
180
180
|
$ agent-network project deploy --property apiKey:sk-xxx --property role:user
|
|
181
181
|
|
|
182
182
|
$ agent-network project deploy --path ./my-project --property myAgent.url:https://api.example.com
|
|
183
|
+
|
|
184
|
+
$ agent-network project deploy --deployment-property brokers:replicas:3
|
|
183
185
|
```
|
|
184
186
|
|
|
187
|
+
Replica count defaults to `1`. Set `--deployment-property brokers:replicas:3`, or use
|
|
188
|
+
`--deployment-properties-file settings.yaml` with numeric YAML:
|
|
189
|
+
|
|
190
|
+
```yaml
|
|
191
|
+
brokers:
|
|
192
|
+
replicas: 3
|
|
193
|
+
```
|
|
194
|
+
|
|
195
|
+
`replicas` must be an integer of at least `1`; quoted YAML strings are rejected.
|
|
196
|
+
It applies to Agent Graph and Mule runtime applications on CloudHub 2.0 and Runtime
|
|
197
|
+
Fabric, for both new deployments and redeployments, not to gateway infrastructure.
|
|
198
|
+
Target capacity and platform quotas still apply; deployment readiness polling does
|
|
199
|
+
not guarantee that every requested replica is running.
|
|
200
|
+
|
|
201
|
+
|
|
185
202
|
## `anypoint-cli-agent-fabric-plugin agent-network project publish`
|
|
186
203
|
|
|
187
204
|
Publish current Agent Network project to Anypoint Platform
|
|
@@ -451,6 +468,14 @@ anypoint-cli-agent-fabric-plugin agent-network project publish
|
|
|
451
468
|
anypoint-cli-agent-fabric-plugin agent-network project deploy
|
|
452
469
|
```
|
|
453
470
|
|
|
471
|
+
## Getting started
|
|
472
|
+
|
|
473
|
+
New to Agent Network projects? See [GETTING_STARTED.md](GETTING_STARTED.md) for a walkthrough of what `agent-network project create` scaffolds — the `exchange.json` descriptor, the `agent-network.yaml` topology (registry, connections, brokers), and the sections of an AgentScript broker file.
|
|
474
|
+
|
|
475
|
+
## Troubleshooting
|
|
476
|
+
|
|
477
|
+
Hitting an error? See [TROUBLESHOOTING.md](TROUBLESHOOTING.md) for a reference of every CLI error code and a guide to common dependent-service failures (Exchange, API Manager, Runtime Fabric, Gateway Manager, Access Management), each with likely causes and fixes.
|
|
478
|
+
|
|
454
479
|
## Development
|
|
455
480
|
|
|
456
481
|
### Prerequisites
|