Control LLM spend and Apex governor exposure for high-traffic Agentforce agents via per-user token budgets and graceful fallback. NOT for API rate-limiting of REST endpoints.

Pre-production security checklist for Agentforce deployments: permission scope, data exposure, authentication, logging. NOT for general Salesforce security review (see security-health-check).

Author and run offline evals for Agentforce agents: fixture format, scoring rubrics, regression baselines, CI integration, prompt-change safety. Use BEFORE every prompt or tool change. Covers multi-turn transcripts, refusal checks, tool-call correctness, grounding accuracy. NOT for online A/B testing (use observability). NOT for general Salesforce test-class patterns (use apex-testing-patterns).

Use when configuring Slack-specific Agentforce capabilities AFTER the basic Slack deployment is complete: enabling Slack-native actions, managing public vs. private action scope, mapping Salesforce-to-Slack user identity, or troubleshooting Slack action failures. Triggers: 'add General Slack Actions topic to agent', 'configure canvas creation for Agentforce in Slack', 'Agentforce private action requires user identity mapping', 'Slack agent cannot send DMs after deployment', 'Agentforce Look Up User action not working in Slack', 'how to map Slack users to Salesforce identities for private actions'. NOT for core Agentforce setup, NOT for basic Slack OAuth installation or DM vs channel-mention mode — those are covered by agentforce/agent-channel-deployment.

Design Agentforce conversations that span multiple turns without losing context: session variable scoping, conversation memory, clarifying-question patterns, topic-to-topic handoff, and the right abstractions for accumulating state across turns. NOT for single-turn agent actions (use agent-actions). NOT for channel-specific conversation UX (use agent-channel-deployment).

Pick the right tool shape for each agent action: Apex invocable vs Flow action vs External Service vs Prompt Template vs Data Cloud retrieval. Covers action selection by use case, argument design for LLM clarity, return-shape contracts, error-surfacing, cost implications, and when to chain tools vs keep a single action. NOT for authoring a specific action (use custom-agent-actions-apex). NOT for topic design (use agent-topic-design).

Use this skill when writing SOQL aggregate queries in Apex — GROUP BY, GROUP BY ROLLUP/CUBE, HAVING, COUNT/SUM/AVG/MIN/MAX, AggregateResult, and date grouping functions. Trigger keywords: soql aggregate groupby apex, count sum group by salesforce, aggregateresult get alias, having clause soql, rollup cube subtotals. NOT for relationship subqueries or inner queries (use apex-soql-relationship-queries), NOT for the Reporting API or Analytics Wave API.

Use this skill when you need to run one Batch Apex job immediately after another completes — chaining via finish(), managing Flex Queue capacity, or choosing between batch-to-batch chaining and a Queueable bridge. NOT for async job technology selection — use the async-selection decision tree. NOT for single-job batch patterns, scope sizing, or Database.Stateful design — use batch-apex-patterns.

Use when building Apex classes meant to be invoked dynamically — from Flow, external packages, managed-package extensions, or loose-coupling code that cannot directly reference the concrete class. Trigger keywords: Callable, call method, dynamic Apex, action registry, plugin pattern, managed package extension point. NOT for: Invocable methods exposed to Flow (see apex-invocable-methods) or REST endpoints (see apex-rest-services).

Apex enum patterns — enum-based dispatch instead of string switching, ordinal stability, enum values in Custom Metadata, package-global enums, and the limitations (no methods on enum constants, no associated data). Covers `Enum.values()`, `valueOf(String)` failures, and the right way to map an enum to/from a picklist value. NOT for picklist-field design (use admin/picklist-design).

Apex Platform Event subscriber runtime semantics — checkpoint-and-resume via `EventBus.TriggerContext.setResumeCheckpoint`, `EventBus.RetryableException` semantics, the 2,000-event-per-trigger default batch (10× standard DML trigger size), and `PlatformEventSubscriberConfig` for batch-size + running-user tuning. Covers the checkpoint-vs-RetryableException decision (the most common subscriber bug). NOT for declaring Platform Events themselves (use platform_events_definition skill), NOT for publishing them (apex/platform-event-publish), NOT for the basic `trigger Foo on MyEvent__e (after insert)` syntax.

Use when serializing Apex objects to JSON strings or deserializing JSON responses into Apex types — especially for callout payloads, integration parsing, and controlling null field output. Trigger keywords: 'suppress null fields JSON Apex', 'deserialize JSON into Apex class', 'JSON parse unknown shape', 'TypeException JSON deserialize', 'JSONGenerator streaming'. NOT for REST endpoint response shaping (use apex-rest-services), NOT for Apex remote actions returning JSON to LWC.

Use this skill when writing Apex that must check governor limits at runtime before executing expensive operations — guard clauses, early-exit patterns, Queueable re-queue on limit approach, and batch scope sizing. Trigger keywords: check governor limits before SOQL apex, defensive coding against limits apex, Limits.getDMLStatements getLimitDMLStatements, Limits class usage, guard clause governor limits, remaining SOQL queries Apex, heap size check before DML, LimitException handling. NOT for limit values themselves — see apex-cpu-and-heap-optimization. NOT for async job design choices — use the async-selection decision tree. NOT for org-level aggregate limit consumption — see architect/org-limits-monitoring.

Grant row-level access programmatically via __Share records when declarative sharing rules cannot express the policy. NOT for OWD, role hierarchy, or criteria-based sharing rule design.

Polymorphic SOQL with TYPEOF: querying Task.WhatId, Task.WhoId, ContentDocumentLink.LinkedEntityId, FeedItem.ParentId; fallback to Type filters; indexing and selectivity. NOT for Activity object model (use activity-and-task-patterns). NOT for general SOQL (use apex-soql-patterns).

Use this skill when you need guaranteed post-Queueable cleanup, retry, or failure-logging logic that must run even when the parent Queueable throws an unhandled exception. Trigger keywords: FinalizerContext, System.attachFinalizer, Queueable cleanup on failure, post-job compensation, guaranteed async cleanup. NOT for batch job completion callbacks — use apex-batch-chaining. NOT for platform event publishing on failure — use platform-events-apex.

Use when designing wrapper or inner classes in Apex to combine SObjects with computed fields, shape data for LWC consumption, or sort collections with Comparable or Comparator. Trigger keywords: wrapper class, inner class, Comparable, Comparator, @AuraEnabled fields, @JsonAccess. NOT for JSON serialization mechanics — use apex-json-serialization. NOT for LWC data-binding patterns — use lwc-reactive-state-patterns or lwc-lightning-record-forms.

Provision and harden integration (API-only) users: no UI login, IP restrictions, minimum permission set, session lifetime, and monitoring. NOT for human admin account hardening.

Composable commerce on Salesforce: headless API layer, micro-frontends, BFF pattern, CDN strategy, third-party composability over B2C/B2B Commerce. NOT for the standard B2C Storefront UX (use b2c-commerce-storefront-setup). NOT for Salesforce Order Management basics (use salesforce-order-management-setup).