This repository treats common Salesforce engineering tasks as modular, executable skills that an AI assistant can load on demand instead of human-copying patterns from docs or templates. That design turns organizational conventions — handler patterns, metadata deploy strategies, sample apps — into consumable units an agent can run, inspect, and update automatically. Expect frequent changes as the skills evolve.
What Sets It Apart
- Skills-as-code packaging: each skill is a directory with a SKILL.md, optional scripts, references, and assets so agents can load metadata (what to trigger) and the implementation (how to run it). This means agents can apply org-specific patterns deterministically rather than guessing from generic docs.
- Multi-tool compatibility: follows the Agent Skills specification, so skills are auto-installed in Agentforce Vibes and usable with OpenCode, Claude Code, Codex, Cursor, and other tools that implement the spec — making the same skill library drive multiple agent runtimes.
- Developer-focused samples & CI sync: includes synced sample apps and GitHub Actions for keeping samples current with upstream npm packages, which reduces drift between examples and production patterns.
- Scope-aware templates: contains focused skills for Apex generation, Flow automation, LWC scaffolding, SOQL queries, permission-set workflows, and safe metadata deployment — designed for common Salesforce developer workflows rather than generic agent tasks.
Who It's For and Trade-offs
Great fit if you manage Salesforce teams and want AI assistants to reliably follow your org's coding patterns, automate routine scaffolding, or run repeatable metadata tasks. It’s also useful for teams adopting Agentforce Vibes or any agent runtime that supports the Agent Skills spec. Look elsewhere if you need stable, long‑term platform guarantees: the repository is explicitly fast‑moving (skills may be renamed, restructured, or removed) and is not a substitute for Salesforce platform APIs or formal SDKs. Also, it assumes the consumer agent/runtime implements the Agent Skills spec — without that, the value drops significantly.
Where It Fits
Use this library as the bridge between Salesforce developer conventions and AI-driven automation: it complements platform APIs and CI/CD tooling by making human patterns machine-consumable. For low-level API stability and long-term integration contracts, rely on official platform APIs and metadata tooling; use sf-skills to encode and automate higher-level team patterns.