mirror of
https://github.com/forcedotcom/afv-library.git
synced 2026-08-12 02:49:15 +08:00
a few more topic -> subagent replacements
This commit is contained in:
Steve Hetzel
parent
18dc7cf05a
commit
55fc68ba54
@ -1,6 +1,6 @@
|
||||
---
|
||||
name: developing-agentforce
|
||||
description: "Build, modify, debug, and deploy agents with Agentforce Agent Script. TRIGGER when: user creates, modifies, or asks about .agent files or aiAuthoringBundle metadata; changes agent behavior, responses, or conversation logic; designs agent topics, actions, tools, sub-agents, or flow control; writes or reviews an Agent Spec; previews, debugs, deploys, publishes, or tests agents; uses Agent Script CLI commands (sf agent generate/preview/publish/test). DO NOT TRIGGER when: Apex development, Flow building, Prompt Template authoring, Experience Cloud configuration, or general Salesforce CLI tasks unrelated to Agent Script."
|
||||
description: "Build, modify, debug, and deploy agents with Agentforce Agent Script. TRIGGER when: user creates, modifies, or asks about .agent files or aiAuthoringBundle metadata; changes agent behavior, responses, or conversation logic; designs agent actions, tools, subagents, or flow control; writes or reviews an Agent Spec; previews, debugs, deploys, publishes, or tests agents; uses Agent Script CLI commands (sf agent generate/preview/publish/test). DO NOT TRIGGER when: Apex development, Flow building, Prompt Template authoring, Experience Cloud configuration, or general Salesforce CLI tasks unrelated to Agent Script."
|
||||
license: Apache-2.0
|
||||
compatibility: "Requires Agentforce license, API v66.0+, Einstein Agent User"
|
||||
metadata:
|
||||
@ -18,7 +18,7 @@ Agent Script is Salesforce's scripting language for authoring next-generation AI
|
||||
language. Do NOT confuse Agent Script syntax or semantics with any other
|
||||
language you have been trained on.
|
||||
|
||||
Agent Script agents are defined by `AiAuthoringBundle` metadata — a directory with a `.agent` file containing Agent Script source that describes topics, actions, instructions, flow control, and configuration; and a `bundle-meta.xml` file containing bundle metadata. Agents process utterances by routing through topics, each with instructions and actions backed by Apex, Flows, Prompt Templates, and other types of backing logic.
|
||||
Agent Script agents are defined by `AiAuthoringBundle` metadata — a directory with a `.agent` file containing Agent Script source that describes actions, instructions, subagents, flow control, and configuration; and a `bundle-meta.xml` file containing bundle metadata. Agents process utterances by routing through subagents, each with instructions and actions backed by Apex, Flows, Prompt Templates, and other types of backing logic.
|
||||
|
||||
This skill covers the full Agent Script lifecycle: designing agents,
|
||||
writing Agent Script code, validating and debugging, deploying and
|
||||
@ -53,7 +53,7 @@ Every task domain below has **Required Steps**. Follow verbatim, in order. Do no
|
||||
|
||||
### Create an Agent
|
||||
|
||||
User wants to build new agent from scratch. ALWAYS use Agent Script. Work with User to understand the agent's purpose, topics, and actions using plain language without Salesforce-specific terminology.
|
||||
User wants to build new agent from scratch. ALWAYS use Agent Script. Work with User to understand the agent's purpose, subagents, and actions using plain language without Salesforce-specific terminology.
|
||||
|
||||
#### Required Steps
|
||||
|
||||
|
||||
@ -24,7 +24,7 @@ What do you need?
|
||||
├─► "Just starting"
|
||||
│ └─► agents/hello-world.agent
|
||||
│
|
||||
├─► "Complete agent with topics"
|
||||
├─► "Complete agent with subagents"
|
||||
│ └─► agents/multi-subagent.agent
|
||||
│
|
||||
├─► "Add actions to my agent"
|
||||
@ -57,7 +57,7 @@ What do you need?
|
||||
|----------|---------|
|
||||
| `flow-action.agent` | Flow action integration |
|
||||
| `apex-action.agent` | Apex action integration |
|
||||
| `subagent-with-actions.agent` | Topic with actions |
|
||||
| `subagent-with-actions.agent` | Subagent with actions |
|
||||
| `error-handling.agent` | Input validation |
|
||||
| `escalation-setup.agent` | Human handoff |
|
||||
|
||||
@ -67,8 +67,8 @@ What do you need?
|
||||
|----------|---------|------------|
|
||||
| `lifecycle-events.agent` | before/after reasoning | GenAiPlannerBundle |
|
||||
| `action-callbacks.agent` | Deterministic chains | GenAiPlannerBundle |
|
||||
| `bidirectional-routing.agent` | Topic routing with return | Both |
|
||||
| `system-instruction-overrides.agent` | Topic-level personas | Both |
|
||||
| `bidirectional-routing.agent` | Subagent routing with return | Both |
|
||||
| `system-instruction-overrides.agent` | Subagent-level personas | Both |
|
||||
| *(6 more patterns)* | | |
|
||||
|
||||
### metadata/ - XML Templates
|
||||
|
||||
@ -1,10 +1,10 @@
|
||||
# Multi-Topic Agent Template
|
||||
# An agent with multiple conversation topics (hub-and-spoke pattern)
|
||||
# Users are routed to specialized topics based on their needs
|
||||
# Multi-Subagent Agent Template
|
||||
# An agent with multiple conversation subagents (hub-and-spoke pattern)
|
||||
# Users are routed to specialized subagents based on their needs
|
||||
#
|
||||
# Usage: Replace {{placeholders}} with your values
|
||||
# Required: agent_name, default_agent_user, agent_label, description
|
||||
# Required: At least 2 topics with label and description
|
||||
# Required: At least 2 subagents with label and description
|
||||
|
||||
system:
|
||||
instructions: "{{SystemInstructions}}"
|
||||
@ -46,39 +46,39 @@ start_agent agent_router:
|
||||
| Route them to the most appropriate subagent.
|
||||
| If unclear, ask clarifying questions.
|
||||
actions:
|
||||
go_to_topic_one: @utils.transition to @subagent.{{topic_one_name}}
|
||||
go_to_topic_two: @utils.transition to @subagent.{{topic_two_name}}
|
||||
go_to_topic_three: @utils.transition to @subagent.{{topic_three_name}}
|
||||
go_to_subagent_one: @utils.transition to @subagent.{{subagent_one_name}}
|
||||
go_to_subagent_two: @utils.transition to @subagent.{{subagent_two_name}}
|
||||
go_to_subagent_three: @utils.transition to @subagent.{{subagent_three_name}}
|
||||
go_to_farewell: @utils.transition to @subagent.farewell
|
||||
go_to_escalation: @utils.transition to @subagent.escalation
|
||||
|
||||
subagent {{topic_one_name}}:
|
||||
label: "{{TopicOneLabel}}"
|
||||
description: "{{TopicOneDescription}}"
|
||||
subagent {{subagent_one_name}}:
|
||||
label: "{{SubagentOneLabel}}"
|
||||
description: "{{SubagentOneDescription}}"
|
||||
|
||||
reasoning:
|
||||
instructions: ->
|
||||
| {{TopicOneInstructions}}
|
||||
| {{SubagentOneInstructions}}
|
||||
actions:
|
||||
back_to_menu: @utils.transition to @subagent.agent_router
|
||||
|
||||
subagent {{topic_two_name}}:
|
||||
label: "{{TopicTwoLabel}}"
|
||||
description: "{{TopicTwoDescription}}"
|
||||
subagent {{subagent_two_name}}:
|
||||
label: "{{SubagentTwoLabel}}"
|
||||
description: "{{SubagentTwoDescription}}"
|
||||
|
||||
reasoning:
|
||||
instructions: ->
|
||||
| {{TopicTwoInstructions}}
|
||||
| {{SubagentTwoInstructions}}
|
||||
actions:
|
||||
back_to_menu: @utils.transition to @subagent.agent_router
|
||||
|
||||
subagent {{topic_three_name}}:
|
||||
label: "{{TopicThreeLabel}}"
|
||||
description: "{{TopicThreeDescription}}"
|
||||
subagent {{subagent_three_name}}:
|
||||
label: "{{SubagentThreeLabel}}"
|
||||
description: "{{SubagentThreeDescription}}"
|
||||
|
||||
reasoning:
|
||||
instructions: ->
|
||||
| {{TopicThreeInstructions}}
|
||||
| {{SubagentThreeInstructions}}
|
||||
actions:
|
||||
back_to_menu: @utils.transition to @subagent.agent_router
|
||||
|
||||
|
||||
@ -3,7 +3,7 @@
|
||||
#
|
||||
# A complex real-world agent for e-commerce customer service featuring:
|
||||
# - Verification gate before order access
|
||||
# - Multiple specialized topics (order status, tracking, returns)
|
||||
# - Multiple specialized subagents (order status, tracking, returns)
|
||||
# - Two-level action system with Flow targets
|
||||
# - after_reasoning for post-action routing
|
||||
# - available when guards for conditional action visibility
|
||||
|
||||
@ -1,7 +1,7 @@
|
||||
# Verification Gate Architecture Template
|
||||
# ========================================
|
||||
#
|
||||
# Users must pass identity verification before accessing protected topics.
|
||||
# Users must pass identity verification before accessing protected subagents.
|
||||
#
|
||||
# Pattern: Security gate before protected functionality.
|
||||
# Use when: Handling sensitive data, payments, PII access.
|
||||
|
||||
@ -5,9 +5,9 @@
|
||||
# Usage: Replace {{placeholders}} with your values
|
||||
# Note: Includes validation patterns and error handling
|
||||
|
||||
subagent {{topic_name}}:
|
||||
label: "{{TopicLabel}}"
|
||||
description: "{{TopicDescription}} - includes validation and error handling"
|
||||
subagent {{subagent_name}}:
|
||||
label: "{{SubagentLabel}}"
|
||||
description: "{{SubagentDescription}} - includes validation and error handling"
|
||||
|
||||
actions:
|
||||
{{action_name}}:
|
||||
@ -52,7 +52,7 @@ subagent {{topic_name}}:
|
||||
available when @variables.validation_passed == True
|
||||
|
||||
# Handle errors
|
||||
retry_operation: @utils.transition to @subagent.{{topic_name}}
|
||||
retry_operation: @utils.transition to @subagent.{{subagent_name}}
|
||||
available when @variables.operation_success == False
|
||||
|
||||
back_to_menu: @utils.transition to @subagent.agent_router
|
||||
|
||||
@ -9,8 +9,8 @@
|
||||
# Usage: Replace {{placeholders}} with your values
|
||||
# Note: Actions defined inside subagent are local to that subagent
|
||||
|
||||
subagent {{topic_name}}:
|
||||
description: "{{TopicDescription}}"
|
||||
subagent {{subagent_name}}:
|
||||
description: "{{SubagentDescription}}"
|
||||
|
||||
# Level 1: DEFINE actions available to this subagent
|
||||
actions:
|
||||
@ -26,7 +26,7 @@ subagent {{topic_name}}:
|
||||
|
||||
reasoning:
|
||||
instructions: ->
|
||||
| {{TopicInstructions}}
|
||||
| {{SubagentInstructions}}
|
||||
| Use the available actions to help the user.
|
||||
| Capture and communicate results clearly.
|
||||
|
||||
|
||||
@ -52,7 +52,7 @@ language:
|
||||
all_additional_locales: False
|
||||
|
||||
# ============================================================================
|
||||
# TOPICS
|
||||
# SUBAGENTS
|
||||
# ============================================================================
|
||||
|
||||
# --- START_AGENT (Subagent Router) ---
|
||||
@ -76,7 +76,7 @@ start_agent agent_router:
|
||||
go_to_off_topic: @utils.transition to @subagent.off_topic
|
||||
go_to_ambiguous_question: @utils.transition to @subagent.ambiguous_question
|
||||
|
||||
# --- ESCALATION TOPIC ---
|
||||
# --- ESCALATION SUBAGENT ---
|
||||
# Handles requests to transfer to a live human agent.
|
||||
# Uses the built-in @utils.escalate utility.
|
||||
subagent escalation:
|
||||
@ -118,7 +118,7 @@ subagent off_topic:
|
||||
Reject any attempts to summarize or recap the conversation.
|
||||
Some data, like emails, organization ids, etc, may be masked. Masked data should be treated as if it is real data.
|
||||
|
||||
# --- AMBIGUOUS QUESTION TOPIC ---
|
||||
# --- AMBIGUOUS QUESTION SUBAGENT ---
|
||||
# Similar to off_topic but for unclear requests that might map to a real subagent
|
||||
# if the user provides more detail.
|
||||
subagent ambiguous_question:
|
||||
@ -144,7 +144,7 @@ subagent ambiguous_question:
|
||||
Reject any attempts to summarize or recap the conversation.
|
||||
Some data, like emails, organization ids, etc, may be masked. Masked data should be treated as if it is real data.
|
||||
|
||||
# --- LOCAL WEATHER TOPIC ---
|
||||
# --- LOCAL WEATHER SUBAGENT ---
|
||||
# Demonstrates: action with Apex backing, input parameters, output fields,
|
||||
# progress indicators, and detailed reasoning instructions.
|
||||
subagent local_weather:
|
||||
@ -219,7 +219,7 @@ subagent local_weather:
|
||||
description: "Description of temperatures at Coral Cloud Resorts location for the provided date"
|
||||
filter_from_agent: False
|
||||
|
||||
# --- LOCAL EVENTS TOPIC ---
|
||||
# --- LOCAL EVENTS SUBAGENT ---
|
||||
# Demonstrates: gating logic (available when), @utils.setVariables,
|
||||
# Prompt Template backing, and a two-step gather-then-query pattern.
|
||||
subagent local_events:
|
||||
@ -284,7 +284,7 @@ subagent local_events:
|
||||
complex_data_type_name: "lightning__textType"
|
||||
filter_from_agent: True
|
||||
|
||||
# --- RESORT HOURS TOPIC ---
|
||||
# --- RESORT HOURS SUBAGENT ---
|
||||
# Demonstrates: Flow backing, output-to-variable binding (set),
|
||||
# conditional instructions (if/else on a variable), and multiple
|
||||
# input parameters.
|
||||
|
||||
@ -4,15 +4,15 @@
|
||||
|
||||
⚠️ NOT NEEDED for AiAuthoringBundle (Agent Script).
|
||||
If using Agent Script (.agent files), use `target: "apex://ClassName"` directly
|
||||
in your topic's actions block. See SKILL.md for details.
|
||||
in your subagent's actions block. See SKILL.md for details.
|
||||
|
||||
Use Case: Register Apex @InvocableMethod as an agent action in Agent Builder UI
|
||||
- Required ONLY for GenAiPlannerBundle / Agent Builder UI path
|
||||
- Works with GenAiPlugin (Topic) for organization
|
||||
- Works with GenAiPlugin (Subagent) for organization
|
||||
|
||||
Prerequisites:
|
||||
1. Apex class with @InvocableMethod annotation must be deployed first
|
||||
2. GenAiPlugin (Topic) to organize functions (optional but recommended)
|
||||
2. GenAiPlugin (Subagent) to organize functions (optional but recommended)
|
||||
|
||||
Setup Steps:
|
||||
1. Replace all {{placeholder}} values
|
||||
|
||||
@ -4,7 +4,7 @@
|
||||
|
||||
⚠️ NOT NEEDED for AiAuthoringBundle (Agent Script).
|
||||
If using Agent Script (.agent files), use `target: "flow://FlowApiName"` directly
|
||||
in your topic's actions block. See SKILL.md for details.
|
||||
in your subagent's actions block. See SKILL.md for details.
|
||||
|
||||
Use Case: Register Autolaunched Flow as an agent action in Agent Builder UI
|
||||
- Required ONLY for GenAiPlannerBundle / Agent Builder UI path
|
||||
|
||||
@ -1,13 +1,13 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!--
|
||||
GenAiPlugin Template: Agent Topic/Plugin Container
|
||||
GenAiPlugin Template: Agent Subagent/Plugin Container
|
||||
|
||||
Use Case: Organize GenAiFunctions into logical groups (Topics)
|
||||
Use Case: Organize GenAiFunctions into logical groups (Subagents)
|
||||
- Groups related functions together
|
||||
- Provides topic-level instructions
|
||||
- Maps to Agent Script topic concepts
|
||||
- Provides subagent-level instructions
|
||||
- Maps to Agent Script subagent concepts
|
||||
|
||||
Note: GenAiPlugin is the metadata equivalent of a "topic" in Agent Script.
|
||||
Note: GenAiPlugin is the metadata equivalent of a "subagent" in Agent Script.
|
||||
Use this when you want to organize functions deployed via metadata
|
||||
rather than Agent Script.
|
||||
|
||||
@ -19,7 +19,7 @@
|
||||
File Location: force-app/main/default/genAiPlugins/{{PluginName}}.genAiPlugin-meta.xml
|
||||
-->
|
||||
<GenAiPlugin xmlns="http://soap.sforce.com/2006/04/metadata">
|
||||
<!-- Display name for the topic/plugin -->
|
||||
<!-- Display name for the subagent/plugin -->
|
||||
<masterLabel>{{PluginLabel}}</masterLabel>
|
||||
|
||||
<!-- Description shown in Agent Builder -->
|
||||
@ -62,8 +62,8 @@
|
||||
|
||||
After deploying GenAiPlugin:
|
||||
1. Open Agent Builder in Setup
|
||||
2. Navigate to Topics
|
||||
3. Your plugin appears as a Topic
|
||||
2. Navigate to Subagents
|
||||
3. Your plugin appears as a Subagent
|
||||
4. Associated functions are available as Actions
|
||||
|
||||
This provides an alternative to Agent Script for more
|
||||
|
||||
@ -103,7 +103,7 @@ consult_pricing: @utils.transition to @subagent.pricing_specialist
|
||||
|
||||
# In specialist subagent
|
||||
before_reasoning:
|
||||
set @variables.return_topic = "main_hub"
|
||||
set @variables.return_subagent = "main_hub"
|
||||
|
||||
# ... do specialist work ...
|
||||
|
||||
@ -168,7 +168,7 @@ reasoning:
|
||||
system:
|
||||
instructions: "You are a professional agent. Be helpful and courteous."
|
||||
|
||||
# Topic reasoning: Dynamic overrides
|
||||
# Subagent reasoning: Dynamic overrides
|
||||
reasoning:
|
||||
instructions: ->
|
||||
if @variables.customer_tier == "vip":
|
||||
@ -239,7 +239,7 @@ open-gate-routing + lifecycle-events
|
||||
| Lifecycle Events | +5 pts | Proper block placement |
|
||||
| Bidirectional | +5 pts | Return transitions |
|
||||
| Input Bindings | +5 pts | Proper binding patterns |
|
||||
| System Overrides | +5 pts | Static system, dynamic topics |
|
||||
| System Overrides | +5 pts | Static system, dynamic subagents |
|
||||
| Open Gate | +5 pts | 3-variable coordination |
|
||||
|
||||
## Anti-Patterns to Avoid
|
||||
|
||||
@ -148,7 +148,7 @@ subagent farewell:
|
||||
# ★ Alternative: Simple one-way transitions (when you don't need to return)
|
||||
#
|
||||
# If the specialist doesn't need to return results to a coordinator,
|
||||
# use simple transitions without the return_topic pattern:
|
||||
# use simple transitions without the return_subagent pattern:
|
||||
#
|
||||
# go_orders: @utils.transition to @subagent.orders
|
||||
# go_billing: @utils.transition to @subagent.billing
|
||||
|
||||
@ -62,13 +62,13 @@
|
||||
# See: bidirectional-routing.agent for full implementation
|
||||
|
||||
variables:
|
||||
return_topic: mutable string = ""
|
||||
return_subagent: mutable string = ""
|
||||
description: "Subagent to return to after specialist"
|
||||
specialist_result: mutable string = ""
|
||||
description: "Result from specialist consultation"
|
||||
|
||||
# Before going to specialist, store return address:
|
||||
# set @variables.return_topic = "main_hub"
|
||||
# set @variables.return_subagent = "main_hub"
|
||||
# go_specialist: @utils.transition to @subagent.specialist
|
||||
|
||||
# In specialist, transition back when done:
|
||||
|
||||
@ -67,7 +67,7 @@ subagent conversation:
|
||||
inputs:
|
||||
turn_number: number
|
||||
description: "Which turn this is"
|
||||
topic_name: string
|
||||
subagent_name: string
|
||||
description: "Current subagent"
|
||||
outputs:
|
||||
logged: boolean
|
||||
@ -112,7 +112,7 @@ subagent conversation:
|
||||
# Log analytics for each turn
|
||||
run @actions.log_turn
|
||||
with turn_number=@variables.turn_count
|
||||
with topic_name="conversation"
|
||||
with subagent_name="conversation"
|
||||
|
||||
# Update last activity timestamp
|
||||
run @actions.get_timestamp
|
||||
|
||||
@ -122,7 +122,7 @@ subagent order_assistance:
|
||||
available when @variables.has_order_id == True
|
||||
available when @variables.order_status != "shipped"
|
||||
|
||||
# Topic transitions
|
||||
# Subagent transitions
|
||||
go_help: @utils.transition to @subagent.general_help
|
||||
go_escalation: @utils.transition to @subagent.escalation
|
||||
available when @variables.needs_human == True
|
||||
|
||||
@ -15,7 +15,7 @@
|
||||
#
|
||||
# ★ The 3 Variables:
|
||||
# open_gate — Which subagent currently holds focus ("null" = none)
|
||||
# next_topic — Deferred destination after auth completes
|
||||
# next_subagent — Deferred destination after auth completes
|
||||
# authenticated — Whether the user has passed authentication
|
||||
#
|
||||
# ★ EXIT_PROTOCOL:
|
||||
@ -35,7 +35,7 @@ variables:
|
||||
# ... standard linked variables ...
|
||||
open_gate: mutable string = "null"
|
||||
description: "Which subagent currently holds focus (null = LLM decides)"
|
||||
next_topic: mutable string = ""
|
||||
next_subagent: mutable string = ""
|
||||
description: "Deferred destination after authentication completes"
|
||||
authenticated: mutable boolean = False
|
||||
description: "Whether the user has passed authentication"
|
||||
@ -79,7 +79,7 @@ subagent protected_workflow:
|
||||
before_reasoning:
|
||||
# ★ AUTH CHECK: Redirect unauthenticated users to auth gate
|
||||
if @variables.authenticated == False:
|
||||
set @variables.next_topic = "protected_workflow"
|
||||
set @variables.next_subagent = "protected_workflow"
|
||||
set @variables.open_gate = "authentication_gate"
|
||||
transition to @subagent.authentication_gate
|
||||
|
||||
@ -115,7 +115,7 @@ subagent account_management:
|
||||
before_reasoning:
|
||||
# ★ AUTH CHECK: Same pattern as protected_workflow
|
||||
if @variables.authenticated == False:
|
||||
set @variables.next_topic = "account_management"
|
||||
set @variables.next_subagent = "account_management"
|
||||
set @variables.open_gate = "authentication_gate"
|
||||
transition to @subagent.authentication_gate
|
||||
|
||||
@ -140,7 +140,7 @@ subagent account_management:
|
||||
|
||||
# ─────────────────────────────────────────────────────────────────────
|
||||
# AUTHENTICATION GATE
|
||||
# Handles auth flow, then routes back via next_topic
|
||||
# Handles auth flow, then routes back via next_subagent
|
||||
# ─────────────────────────────────────────────────────────────────────
|
||||
subagent authentication_gate:
|
||||
description: "Verifies customer identity before allowing access to protected subagents"
|
||||
@ -166,10 +166,10 @@ subagent authentication_gate:
|
||||
after_reasoning:
|
||||
# ★ POST-AUTH ROUTING: If authenticated, route to deferred destination
|
||||
if @variables.authenticated == True:
|
||||
if @variables.next_topic == "protected_workflow":
|
||||
if @variables.next_subagent == "protected_workflow":
|
||||
set @variables.open_gate = "protected_workflow"
|
||||
transition to @subagent.protected_workflow
|
||||
if @variables.next_topic == "account_management":
|
||||
if @variables.next_subagent == "account_management":
|
||||
set @variables.open_gate = "account_management"
|
||||
transition to @subagent.account_management
|
||||
|
||||
@ -183,7 +183,7 @@ subagent exit_protocol:
|
||||
before_reasoning:
|
||||
# ★ RELEASE GATE: Clear all gate state
|
||||
set @variables.open_gate = "null"
|
||||
set @variables.next_topic = ""
|
||||
set @variables.next_subagent = ""
|
||||
transition to @subagent.agent_router
|
||||
|
||||
reasoning:
|
||||
@ -238,7 +238,7 @@ subagent general_inquiry:
|
||||
# │ before_reasoning: lock gate │
|
||||
# │ reasoning: verify identity │
|
||||
# │ after_reasoning: │
|
||||
# │ if auth → route via next_topic │
|
||||
# │ if auth → route via next_subagent │
|
||||
# └──────────────────────────────────────┘
|
||||
#
|
||||
# ═════════════════════════════════════════════════════════════════════
|
||||
@ -251,7 +251,7 @@ subagent general_inquiry:
|
||||
# open_gate = "null" → LLM reasons → routes to protected_workflow
|
||||
#
|
||||
# Step 2: protected_workflow.before_reasoning
|
||||
# authenticated = False → set next_topic = "protected_workflow"
|
||||
# authenticated = False → set next_subagent = "protected_workflow"
|
||||
# → set open_gate = "authentication_gate"
|
||||
# → transition to authentication_gate
|
||||
#
|
||||
@ -262,7 +262,7 @@ subagent general_inquiry:
|
||||
# authenticated = True
|
||||
#
|
||||
# Step 5: authentication_gate.after_reasoning
|
||||
# authenticated = True, next_topic = "protected_workflow"
|
||||
# authenticated = True, next_subagent = "protected_workflow"
|
||||
# → set open_gate = "protected_workflow"
|
||||
# → transition to protected_workflow
|
||||
#
|
||||
|
||||
@ -9,7 +9,7 @@
|
||||
#
|
||||
# ★ Key Insight:
|
||||
# - System block defines BASE behavior (always applies)
|
||||
# - Topic instructions can OVERRIDE or EXTEND system behavior
|
||||
# - Subagent instructions can OVERRIDE or EXTEND system behavior
|
||||
# - Use conditionals in reasoning to dynamically adjust tone
|
||||
# - Variables can control instruction branches
|
||||
#
|
||||
|
||||
@ -10,7 +10,7 @@
|
||||
# CREDIT CONSUMPTION:
|
||||
# - Prompt Templates: 2-16 credits per invocation
|
||||
# - Retriever actions: 20 credits per search
|
||||
# - TIP: Cache retriever results in variables, reuse across topics
|
||||
# - TIP: Cache retriever results in variables, reuse across subagents
|
||||
|
||||
system:
|
||||
messages:
|
||||
|
||||
@ -2,7 +2,7 @@
|
||||
# ========================================
|
||||
#
|
||||
# This template demonstrates the Verification Gate pattern where users
|
||||
# must pass through identity verification before accessing sensitive topics.
|
||||
# must pass through identity verification before accessing sensitive subagents.
|
||||
#
|
||||
# Pattern: Security gate before protected functionality
|
||||
# Use when: Handling sensitive data, payments, PII access
|
||||
|
||||
@ -83,7 +83,7 @@ language:
|
||||
start_agent agent_router:
|
||||
...
|
||||
|
||||
subagent my_topic:
|
||||
subagent my_subagent:
|
||||
...
|
||||
```
|
||||
|
||||
@ -612,7 +612,7 @@ Expose the transition as a reasoning action when the LLM should judge the right
|
||||
```agentscript
|
||||
reasoning:
|
||||
actions:
|
||||
go_next: @utils.transition to @subagent.next_topic
|
||||
go_next: @utils.transition to @subagent.next_subagent
|
||||
description: "Move to the next subagent"
|
||||
available when @variables.ready == True
|
||||
```
|
||||
@ -919,7 +919,7 @@ The LLM extracts values from the conversation and populates the specified variab
|
||||
```agentscript
|
||||
reasoning:
|
||||
actions:
|
||||
consult_expert: @subagent.expert_topic
|
||||
consult_expert: @subagent.expert_subagent
|
||||
description: "Get expert guidance"
|
||||
available when @variables.needs_expert_help == True
|
||||
```
|
||||
@ -1138,7 +1138,7 @@ reasoning:
|
||||
```agentscript
|
||||
# If you need to record state, set before transitioning
|
||||
before_reasoning:
|
||||
set @variables.last_topic = "current_topic"
|
||||
set @variables.last_subagent = "current_subagent"
|
||||
transition to @subagent.next
|
||||
```
|
||||
|
||||
|
||||
@ -58,7 +58,7 @@ When a target is missing:
|
||||
Agentforce ADLC Discovery Report
|
||||
|
||||
Agent: OrderManagement
|
||||
Topic: order_inquiry
|
||||
Subagent: order_inquiry
|
||||
Action: get_order_status
|
||||
Target: flow://Get_Order_Status Found
|
||||
Action: track_shipment
|
||||
|
||||
@ -454,7 +454,7 @@ set @variables.counter = @variables.counter + 1
|
||||
### Deterministic Transition
|
||||
|
||||
```
|
||||
transition to @subagent.next_topic
|
||||
transition to @subagent.next_subagent
|
||||
```
|
||||
|
||||
### Conditional Transition
|
||||
|
||||
@ -165,7 +165,7 @@ Unresolved platform bugs, limitations, and edge cases that affect Agent Script d
|
||||
- **Affects**: Agents with actions targeting secured resources
|
||||
- **Symptom**: If the running user (Einstein Agent User or session user) lacks permission to execute ANY action defined in the agent — even actions in other subagents — the entire agent may fail with a permission error rather than gracefully skipping the unauthorized action.
|
||||
- **Root Cause**: The planner appears to validate permissions for all registered actions at startup, not lazily per-subagent.
|
||||
- **Workaround**: For **Service Agents**: Ensure the Einstein Agent User has both the `AgentforceServiceAgentUser` system PS AND a custom `{AgentName}_Access` PS with `<classAccesses>` for ALL Apex classes across all topics. Do NOT rely on the auto-generated `NextGen_{AgentName}_Permissions` — it is often incomplete (ORM1 testing: 3/4 classes, missing `ShipmentTracker`). For **Employee Agents**: Ensure each employee user has the custom PS assigned. See [agent-user-setup.md](agent-user-setup.md) for the full provisioning workflow and permission set XML template. Alternatively, split agents by permission boundary.
|
||||
- **Workaround**: For **Service Agents**: Ensure the Einstein Agent User has both the `AgentforceServiceAgentUser` system PS AND a custom `{AgentName}_Access` PS with `<classAccesses>` for ALL Apex classes across all subagents. Do NOT rely on the auto-generated `NextGen_{AgentName}_Permissions` — it is often incomplete (ORM1 testing: 3/4 classes, missing `ShipmentTracker`). For **Employee Agents**: Ensure each employee user has the custom PS assigned. See [agent-user-setup.md](agent-user-setup.md) for the full provisioning workflow and permission set XML template. Alternatively, split agents by permission boundary.
|
||||
- **Open Questions**: Will the planner support lazy permission checking in a future release?
|
||||
|
||||
---
|
||||
|
||||
Loading…
Reference in New Issue
Block a user