salesforce-functions-replacement
Salesforce Functions is retired (EOL Jan 2025). This skill maps Functions workloads to replacements: Heroku (with Hyperforce), external containers, Apex (where viable), Agentforce Actions, external compute via Named Credentials. NOT for Lambda / Azure Functions tutorials. NOT for Apex @future replacement (use async-selection tree).
Best use case
salesforce-functions-replacement is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Salesforce Functions is retired (EOL Jan 2025). This skill maps Functions workloads to replacements: Heroku (with Hyperforce), external containers, Apex (where viable), Agentforce Actions, external compute via Named Credentials. NOT for Lambda / Azure Functions tutorials. NOT for Apex @future replacement (use async-selection tree).
Teams using salesforce-functions-replacement should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/salesforce-functions-replacement/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How salesforce-functions-replacement Compares
| Feature / Agent | salesforce-functions-replacement | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Salesforce Functions is retired (EOL Jan 2025). This skill maps Functions workloads to replacements: Heroku (with Hyperforce), external containers, Apex (where viable), Agentforce Actions, external compute via Named Credentials. NOT for Lambda / Azure Functions tutorials. NOT for Apex @future replacement (use async-selection tree).
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Salesforce Functions Replacement
Activate when migrating off Salesforce Functions (end-of-life January 2025). Every Functions workload needs a new home, and the right home depends on the workload. Do not assume "one replacement" — Heroku suits long-running; Apex suits simple compute; external containers suit CPU-heavy work. Agentforce Actions are a good fit for LLM-adjacent logic.
## Before Starting
- **Inventory every Function.** List runtime, invocation pattern, average compute time, data dependencies.
- **Classify by driver.** Why was Functions chosen? Apex-unfriendly library? CPU intensity? Language preference? Each reason maps to a different replacement.
- **Set a migration deadline.** Functions stopped accepting new deploys in late 2024; runtime support ends January 2025. Past EOL, any outage is terminal.
## Core Concepts
### Why Functions existed
Functions offered a managed Node.js / Java / Python runtime tightly integrated with Apex, invoked via `Function.get('ns.name').invoke(payload)`. It filled the gap between Apex limits and external compute without needing to manage a separate platform.
### Heroku as default replacement
Heroku is Salesforce-owned and integrates natively with Hyperforce. It offers Private Spaces, Shield, add-ons. For long-running or language-dependent workloads that Apex can't run, Heroku is the most natural migration target.
### Container-on-Hyperforce patterns
Run containers (AKS, EKS, GKE) or Cloud Run with OAuth-authenticated callouts to Salesforce. Integrate via Named Credentials + Pub/Sub or Platform Events. More ops burden than Heroku; more flexibility.
### Apex (where viable)
Simple compute that fits within Apex limits can move to Apex. CPU-bound Apex is painful (10s CPU limit on sync, 60s on async). Library-dependent work (PDF manipulation, heavy image processing) does not fit Apex.
### Agentforce Actions
LLM / AI-adjacent Functions are a good fit for Agentforce Actions or Agentforce prompt templates. Natural migration for AI-shaped workloads.
## Common Patterns
### Pattern: PDF generation Function → Heroku service
Heroku Node dyno running the same PDF library. Apex calls via Named Credential + REST; Heroku returns the generated PDF. Private Space for PCI/PII workloads.
### Pattern: Heavy compute batch → Heroku + Redis queue
Instead of an Apex-triggered Function, publish a Platform Event → Heroku worker processes job → writes result back via REST. Decouples Apex from long compute.
### Pattern: LLM enrichment → Agentforce Action
Instead of a Function calling OpenAI, define an Agentforce prompt template + Action. Native trust layer, no Heroku dyno to manage.
### Pattern: Simple enrichment → Apex Queueable
If the Function was < 5s of compute and used only standard libraries, rewrite as Queueable Apex and call via `System.enqueueJob`.
## Decision Guidance
| Workload profile | Replacement | Reason |
|---|---|---|
| Long-running, language-specific | Heroku | Native Salesforce path |
| CPU-bound batch | Heroku worker + queue | Decoupled |
| Simple enrichment, <5s | Apex Queueable | Cheapest |
| LLM / AI enrichment | Agentforce Action | Native trust layer |
| Existing container workload elsewhere | Cloud Run / AKS | Leverage existing infra |
## Recommended Workflow
1. Inventory every Function: runtime, calls/day, avg latency, data returned.
2. Classify each Function by driver (language, CPU, library, AI).
3. Assign replacement per Function per decision guidance.
4. Design integration for each: Named Credential + REST, Platform Events, Pub/Sub, Agentforce Action.
5. Build one pilot replacement end-to-end; measure latency and cost.
6. Cut over Function-by-Function; never big-bang.
7. Decommission Functions as each workload moves; track until zero invocations.
## Review Checklist
- [ ] Function inventory captured with drivers
- [ ] Replacement assigned per workload with rationale
- [ ] Named Credentials + OAuth external client apps configured
- [ ] Pilot measured against current Function perf
- [ ] Cost comparison documented
- [ ] Cutover runbook with rollback per workload
- [ ] Functions decommission tracker live
## Salesforce-Specific Gotchas
1. **Heroku Private Space + Shield is required for PCI/PII workloads.** A standard dyno does not meet compliance for protected workloads.
2. **Named Credentials and External Client Apps are the modern auth surface.** Legacy Auth Providers are being deprecated; migrate to External Client Apps during Functions migration.
3. **Pub/Sub API replaces CometD for high-volume eventing.** If the Function was consuming Streaming API, the replacement should use Pub/Sub API.
## Output Artifacts
| Artifact | Description |
|---|---|
| Function inventory | Runtime, driver, replacement target |
| Replacement design per workload | Pattern, endpoints, auth |
| Cost + ops comparison | Functions vs replacement |
| Migration runbook | Per-workload cutover + rollback |
## Related Skills
- `integration/integration-pattern-selection` — integration mechanism choice
- `integration/named-credentials-and-external-auth` — auth to Heroku / external
- `agentforce/agent-action-design` — AI-adjacent migrationsRelated Skills
salesforce-shield-deployment
Roll out Shield (Platform Encryption + Event Monitoring + Field Audit Trail) end-to-end, sequencing feature enablement to avoid data lockout. NOT for Classic Encryption or general PE design.
ferpa-compliance-in-salesforce
Use this skill when implementing FERPA (Family Educational Rights and Privacy Act) compliance controls in Salesforce Education Cloud or Education Data Architecture (EDA): LearnerProfile FERPA boolean fields, directory information opt-out via FLS and Individual data privacy flags, ContactPointTypeConsent for parental and third-party disclosure, 45-day student records response window tracking, and consent workflow automation. Trigger keywords: FERPA, student records privacy, LearnerProfile, parental disclosure, directory information opt-out, education data privacy, student consent, education cloud compliance. NOT for GDPR/CCPA general data privacy (see gdpr-data-privacy skill), platform encryption at rest (see platform-encryption skill), or HIPAA health-data compliance.
industries-cpq-vs-salesforce-cpq
Use this skill when comparing Industries CPQ (formerly Vlocity CPQ) with Salesforce CPQ (Revenue Cloud managed package) — covering feature parity, decision criteria, migration paths, and coexistence patterns. Trigger keywords: Vlocity CPQ, Industries CPQ, Salesforce CPQ comparison, Revenue Cloud migration, CPQ selection, which CPQ to use. NOT for implementing, configuring, or debugging either CPQ product.
tableau-salesforce-connector
Tableau ↔ Salesforce integration patterns: Tableau Salesforce connector, Tableau for Salesforce, CRM Analytics alternative, Data Cloud + Tableau, embedded Tableau dashboards. Choose between connector modes (live, extract, direct-to-Data-Cloud). NOT for CRM Analytics Studio (use crm-analytics-foundation). NOT for generic Tableau Server setup.
slack-salesforce-integration-setup
Use this skill when setting up or troubleshooting the Salesforce for Slack managed app — including connecting a Salesforce org to a Slack workspace, configuring the three-party admin handshake, linking Slack channels to Salesforce records, enabling record preview sharing, and managing org-level limits. Triggers on: Salesforce for Slack app not connecting, Slack org connection setup, Salesforce record sharing in Slack, Slack workspace admin approval, connecting Salesforce to Slack. NOT for building custom Slack apps or Slack bots (separate development platform), not for Slack Workflow Builder Salesforce connector (use slack-workflow-builder skill), not for Flow-based Slack messaging (use flow-for-slack skill).
salesforce-to-salesforce-integration
Use this skill to implement Salesforce-to-Salesforce integration patterns — covering the native S2S feature, API-based cross-org sync, Platform Event bridging, and Salesforce Connect Cross-Org adapter. Trigger keywords: Salesforce to Salesforce integration, cross-org data sharing, S2S feature, cross-org Platform Events, Salesforce Connect cross-org. NOT for multi-org strategy or architecture decisions (use architect/multi-org-strategy), single-org data sharing, or external (non-Salesforce) system integration.
salesforce-maps-setup
Use when configuring Salesforce Maps (formerly MapAnything) — territory planning, route optimization, live tracking, geo-grid visualizations, and check-in/check-out workflows for Sales or Service field reps not on Field Service. Covers package installation order (Maps + Maps Advanced + Maps Routing/Live Tracking add-ons), the MapsTerritoryPlan / MapsAdvancedRoute / MapsLayer object family, base-data syncs (Geocoding and Routing services), and integration with Sales and Service Cloud records. Triggers: 'Salesforce Maps setup', 'MapAnything migration', 'territory planning by polygon', 'route optimization for sales reps', 'live tracking field reps', 'plot accounts on a map', 'check-in to the closest account'. NOT for Field Service Lightning territory and scheduling (use admin/fsl-scheduling-optimization-design and data/fsl-territory-data-setup) — Maps and FSL are different products. NOT for Consumer Goods Cloud retail visit planning (use admin/consumer-goods-cloud-setup) — RoutePlan/Visit objects are CG-specific. NOT for Tableau / CRM Analytics geo charts.
salesforce-data-pipeline-etl
Export large Salesforce datasets to a lakehouse via Bulk API 2.0, CDC streams, or Salesforce Data Pipelines. NOT for ad-hoc exports.
salesforce-connect-external-objects
Use when deciding whether Salesforce Connect and External Objects are the right fit for external data access, or when reviewing OData, cross-org, and custom adapter patterns, query limitations, and latency tradeoffs. Triggers: 'Salesforce Connect', 'External Objects', '__x', 'OData adapter', 'custom adapter'. NOT for ordinary ETL or replicated-data designs where the data should live inside Salesforce.
outbound-webhook-from-salesforce
Use when Salesforce must POST a webhook to a third-party endpoint after a record change — with signed payloads, retries, dead-lettering, rate limits, and idempotency. Covers design choice between Outbound Message, Flow HTTP Callout, Apex Queueable callout, and Event Relay. Does NOT cover inbound webhooks into Salesforce (see inbound-webhook or apex-rest-webhook).
mulesoft-salesforce-connector
Designing and configuring MuleSoft Anypoint Salesforce Connector flows: API selection (SOAP/REST/Bulk/Streaming), OAuth 2.0 JWT Bearer auth, watermark-based incremental sync with Object Store, batch processing with record-level error isolation, and replay topic subscriptions. Use when building Mule 4 flows that read from or write to Salesforce, migrating from Mule 3 watermark to Mule 4 Object Store, or troubleshooting connector authentication and API limits. NOT for native Salesforce-to-Salesforce integration without MuleSoft (use platform-events-integration or change-data-capture-integration). NOT for generic REST callout patterns from Apex (use rest-api-patterns).
heroku-salesforce-integration
Heroku ↔ Salesforce integration paths — Heroku Connect (Postgres bidirectional sync), Heroku AppLink (expose Heroku APIs to Flow / Apex / Agentforce), Heroku External Objects (Salesforce Connect oData), Platform Events, REST API in either direction, and Salesforce Canvas for UI embedding. Decision matrix + Heroku Connect deep dive (OAuth integration user, plan-tier row limits, polling cadence, region co-location). NOT for AWS integration (see integration/aws-salesforce-patterns), NOT for generic PaaS, NOT for MuleSoft.