salesforce-claude-code
CLAUDE.md15 starsChanged 7 months agoMITView on GitHub
Salesforce Claude Code (SCC) is an agent harness performance system — a Claude Code plugin providing production-ready agents, skills, hooks, and MCP configurations specifically for Salesforce development. Works across Claude Code and Cursor. The project is a plugin system — mostly Markdown/JSON content consumed by AI agent harnesses, backed by Node.js scripts for installation, hooks, and validation.
- Installs packages
In this repository58 other files it gives its agents.
- asideSkill
- checkpointSkill
- configure-sccSkill
- continuous-agent-loopSkill
- mcp-server-patternsSkill
- model-routeSkill
- prompt-optimizerSkill
- refactor-cleanSkill
- resume-sessionSkill
- save-sessionSkill
- search-firstSkill
- security-scanSkill
- sessionsSkill
- sf-2gp-security-reviewSkill
- sf-agentforce-developmentSkill
- sf-agentforce-mcp-actionsSkill
- sf-apex-async-patternsSkill
- sf-apex-best-practicesSkill
- sf-apex-constraintsSkill
- sf-apex-cursorSkill
- sf-apex-enterprise-patternsSkill
- sf-apex-testingSkill
- sf-api-designSkill
- sf-approval-processesSkill
- sf-aura-developmentSkill
- sf-build-fixSkill
- sf-cli-referenceSkill
- sf-data-modelingSkill
- sf-debuggingSkill
- sf-deployment-constraintsSkill
- sf-deploymentSkill
- sf-devops-ci-cdSkill
- sf-docs-lookupSkill
- sf-e2e-testingSkill
- sf-experience-cloudSkill
- sf-flow-developmentSkill
- sf-governor-limitsSkill
- sf-harness-auditSkill
- sf-helpSkill
- sf-integrationSkill
- sf-lwc-constraintsSkill
- sf-lwc-developmentSkill
- sf-lwc-testingSkill
- sf-metadata-managementSkill
- sf-platform-events-cdcSkill
- sf-quickstartSkill
- sf-security-constraintsSkill
- sf-securitySkill
- sf-soql-constraintsSkill
- sf-soql-optimizationSkill
- sf-tdd-workflowSkill
- sf-testing-constraintsSkill
- sf-trigger-constraintsSkill
- sf-trigger-frameworksSkill
- sf-visualforce-developmentSkill
- strategic-compactSkill
- update-docsSkill
- update-platform-docsSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 109
Target Salesforce API version: 66.0 (Spring '26): When referencing API-version-specific features (SOQL Cursors, RunRelevantTests, …
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Project Overview
Salesforce Claude Code (SCC) is an agent harness performance system — a Claude Code plugin providing production-ready agents, skills, hooks, and MCP configurations specifically for Salesforce development. Works across Claude Code and Cursor.
- npm package:
scc-universal(CLI install name) - Plugin name:
salesforce-claude-code(marketplace display name)
Build & Test Commands
# Install dependencies
npm install
# Build all derived content (Cursor adapters + MCP copies)
npm run build
# Run full validation + tests (runs build first automatically)
npm test
# Run only unit/integration tests
node tests/run-all.js
# Run individual test files
node tests/lib/utils.test.js
node tests/hooks/hooks.test.js
# Run individual CI validators
node scripts/ci/validate-agents.js
node scripts/ci/validate-skills.js
node scripts/ci/validate-hooks.js
node scripts/ci/validate-install-manifests.js
node scripts/ci/validate-no-personal-paths.js
# Lint (ESLint + markdownlint)
npm run lint
# Auto-fix markdownlint errors
npx markdownlint '**/*.md' --ignore node_modules --fix
# Coverage (80% lines/functions/branches/statements required)
npm run coverage
# Smoke test (tests npm pack + install/uninstall in a temp project)
bash scripts/ci/smoke-test.sh
# SCC CLI
npx scc-universal install <target> # Install SCC content (apex, lwc, all)
npx scc-universal doctor # Diagnose missing/drifted files
npx scc-universal repair # Restore drifted files
npx scc-universal status # Query JSON state store
npx scc-universal uninstall # Remove SCC-managed files
npx scc-universal plan # Preview files to be installed (dry run)
npx scc-universal list-installed # Show currently installed SCC filesArchitecture
The project is a plugin system — mostly Markdown/JSON content consumed by AI agent harnesses, backed by Node.js scripts for installation, hooks, and validation.
Core Content (Markdown + YAML frontmatter)
- agents/ — 17 agents (11 Salesforce domain + 6 platform). Lean workflow orchestrators (~80-150 lines). Preload constraint skills via
skillsfrontmatter. Format: Markdown with YAML frontmatter (name,description,tools,model,origin,readonly,skills). - skills/ — 57 domain-knowledge modules (7 constraint, 29 pattern, 5 utility, 16 platform). Pattern skills are model-invocable (agents read them on demand). Format:
skills/<name>/SKILL.mdwith YAML frontmatter (name,description,origin,user-invocable). - hooks/hooks.json — Claude Code hook lifecycle (SessionStart, PreToolUse, PostToolUse, PostToolUseFailure, PreCompact, Stop, SessionEnd). Hook scripts live in
scripts/hooks/.
Generated Content (Never Edit Directly)
- .cursor/ — Auto-generated from
agents/andskills/bynpm run build. The pre-commit hook enforces sync. Always edit the source inagents/orskills/, then runnpm run buildand stage.cursor/.
Script Infrastructure (Node.js, CommonJS)
- scripts/scc.js — Main CLI entry point (
npx scc-universal). - scripts/cli/ — End-user CLI commands (install, uninstall).
- scripts/dev/ — Contributor/dev tools (doctor, repair, status, build-cursor, harness-audit).
- scripts/lib/ — Shared utilities (package-manager detection, install executor, JSON state store, adapters).
- scripts/hooks/ — Hook implementations. Uses
run-with-flags.jsfor profile-based gating. - scripts/ci/ — CI validators enforcing structure/frontmatter of all content types.
- manifests/ — Selective-install manifests defining what files to install per language/target.
- schemas/ — JSON Schemas for hooks, plugin config, install state, package-manager config.
Cross-Harness Support
- .claude-plugin/ — Plugin manifest for Claude Code marketplace.
- .cursor-plugin/ — Plugin manifest for Cursor marketplace.
Pre-commit Hook
The pre-commit hook (.githooks/pre-commit) mirrors CI locally: build, ESLint, markdownlint, all 6 validators, and full test suite. Setup for new contributors:
git config core.hooksPath .githooksRelease Process
Versioning is automated via release-please. On merge to main:
- Release-please creates a Release PR bumping
package.jsonand all 4 plugin manifests (.claude-plugin/plugin.json,.claude-plugin/marketplace.json,.cursor-plugin/plugin.json,.cursor-plugin/marketplace.json). - Merging the Release PR triggers
npm publish --provenance --access public.
Conventional commit types drive version bumps: feat: → minor, fix: → patch, feat!: → major.
Key Conventions
- Target Salesforce API version: 66.0 (Spring '26). When referencing API-version-specific features (SOQL Cursors, RunRelevantTests, @testFor), always note the minimum version required.
- Node.js >= 20 required. Pinned: Node 20.19.0, Python 3.12.8 (
.tool-versions). - CommonJS throughout — all scripts use
require()/module.exports. - File naming — lowercase with hyphens (e.g.,
sf-apex-agent.md,sf-trigger-frameworks). - Hook profiles —
SCC_HOOK_PROFILE=minimal|standard|strictcontrols which hooks run.SCC_DISABLED_HOOKS=...disables specific hooks. - Package manager — auto-detects npm/pnpm/yarn/bun. Override via
CLAUDE_PACKAGE_MANAGERenv var. - Commit format — Conventional commits:
<type>: <description>. Max header: 100 chars. Types:feat,fix,docs,style,refactor,perf,test,chore,ci,build,revert.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

