forcedotcom/sf-skills / plugins/builder/agentforce-adlc/skills/agentforce-generate
agentforce-generate
FlaggedBuild, modify, audit, repair, optimize, debug, and deploy agents with Agentforce Agent Script. TRIGGER when: user creates, reviews, or changes .agent files or aiAuthoringBundle metadata; asks to fix AgentScript, audit an existing agent, run an AgentScript health check, common-pitfall review, or baseline-versus-candidate repair loop; changes a response, action, subagent, route, state flow, or Agent Spec; migrates or upgrades a legacy GenAiPlannerBundle Agentforce agent (NOT an Einstein Bot) to an aiAuthoringBundle (Agent Script) agent; previews, debugs, deploys, publishes, or tests agents; uses sf agent generate/preview/publish/test; or manages Agentforce MCP servers, tools, assets, or authentication. DO NOT TRIGGER when: Apex, Flow, Prompt Template, Experience Cloud, or general Salesforce CLI work is unrelated to Agent Script; the source is an Einstein Bot (BotDefinition/BotVersion) — use agentforce-bot-upgrade; or the primary input is a production session or trace ID rather than an agent artifact.
assets/agents/router-first.agent
View on GitHub# Router-First Architecture Template
#
# Use a central router when distinct request types need different objectives,
# instructions, actions, authority, or escalation behavior. Do not persist
# ordinary user inputs merely to keep a request handler focused; each action
# uses slot filling from the current turn and surviving conversation history.
system:
messages:
welcome: "Welcome! I can help with orders, returns, or general support."
error: "I apologize, something went wrong. Let me try again."
instructions: |
You are a customer service agent for an e-commerce company. During
request classification, choose exactly one matching request handler
without answering the underlying request. In order and return handlers,
use the available read actions for external facts. In general support,
answer the question or transfer to a human when needed.
access:
default_agent_user: "agent@yourorg.com"
config:
developer_name: "Router_First_Agent"
agent_label: "Customer Service Agent"
description: "Routes orders, returns, and support requests to distinct handlers"
language:
default_locale: "en_US"
start_agent agent_router:
label: "Request Router"
description: "Routes the user's latest request without answering it"
model_config:
model: "model://sfdc_ai__DefaultEinsteinHyperClassifier"
reasoning:
instructions: ->
| Route only the user's latest request. Choose exactly one route,
and do not answer the underlying request in this router.
actions:
check_order: @utils.transition to @subagent.orders
description: "Handle order status or tracking"
handle_return: @utils.transition to @subagent.returns
description: "Handle return eligibility or an existing return"
general_help: @utils.transition to @subagent.support
description: "Handle a general support question"
subagent orders:
label: "Order Help"
description: "Handles order status and tracking inquiries"
actions:
get_order_status:
description: "Look up the current status of a specific order rather than answer a general shipping-policy question"
inputs:
order_id: string
description: "Order number supplied by the customer"
is_required: True
outputs:
status: string
description: "Current order status"
target: "flow://Get_Order_Status"
reasoning:
instructions: ->
| Help with the user's latest order status or tracking request. Ask
for an order number only when it is absent from the current turn
and surviving conversation history. Report only action-returned
status as an external fact.
| If the user changes intent, use the matching transition.
actions:
lookup_order: @actions.get_order_status
description: "Look up a concrete order using the latest order number in the conversation"
with order_id=...
to_returns: @utils.transition to @subagent.returns
description: "Handle a changed request about returns"
reclassify: @utils.transition to @subagent.agent_router
description: "Reclassify any other changed intent"
subagent returns:
label: "Return Help"
description: "Explains return eligibility and looks up existing return status"
actions:
get_return_status:
description: "Look up an already-created return; this does not create a return or issue money"
inputs:
return_id: string
description: "Existing return identifier supplied by the customer"
is_required: True
outputs:
status: string
description: "Current return status"
target: "flow://Get_Return_Status"
reasoning:
instructions: ->
| Explain return eligibility and policy, or look up an existing
return when the user provides a return identifier. This template
intentionally does not create returns or issue refunds; those
consequential actions require machine-checkable authorization,
exact targets, confirmation, idempotency, and failure handling.
| If the user changes intent, return to request routing.
actions:
lookup_return: @actions.get_return_status
description: "Look up an existing return using the latest return identifier in the conversation"
with return_id=...
reclassify: @utils.transition to @subagent.agent_router
description: "Reclassify a changed intent"
subagent support:
label: "General Support"
description: "Handles general support questions that need no order or return action"
reasoning:
instructions: ->
| Answer the user's latest general support question. If the request
requires order or return handling, route to the matching handler.
actions:
go_orders: @utils.transition to @subagent.orders
description: "Handle a changed request about an order"
go_returns: @utils.transition to @subagent.returns
description: "Handle a changed request about a return"
escalate: @utils.escalate
description: "Transfer to a human when the user explicitly requests one or the issue cannot be handled safely"
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |