← ClaudeAtlas

sf-orgdifflisted

Compares two connected Salesforce orgs through their MCP connectors — inventory-first metadata diffing across Apex, Flows, objects/fields, validation rules, permission sets, layouts, labels, endpoints, and installed packages — producing a drift report with severity flags and a reconciliation plan. Strictly read-only on both orgs. Use when the user asks "what drifted", "compare orgs", "sandbox vs production", "did the release deploy everything", "compare against baseline", pre/post-release verification, or any question about differences between two orgs. Do NOT use for auditing a single org's quality (use sf-audit), creating or deploying metadata (use sf-metadata), or vulnerability scanning (use sf-security). Usage: /sf-orgdiff [drift|release|baseline] {source-connector} {target-connector} [scope] ...
kugamon/salesforce-core-skills · ★ 0 · AI & Automation · score 68
Install: claude install-skill kugamon/salesforce-core-skills
# Salesforce Org-to-Org Diff Comparison specialist for two live Salesforce orgs reached through two MCP connectors. Inventory both sides cheaply, diff the inventories, fetch bodies only for the narrow shortlist that actually differs, and hand back a drift report that says what changed, which direction, and what to do about it. **READ-ONLY, always, on both orgs.** This skill never writes to either org — no DML, no metadata deploys, no Tooling POSTs. Reconciliation is a *plan* handed to sf-metadata, never an action taken here. There is no mode or argument that relaxes this. ## Dispatch | First argument or intent | Workflow | | ----------------------------------------------------------------- | -------- | | `drift`, "what drifted", "sandbox vs production", "out of sync" | Drift Check | | `release`, "did the release deploy everything", "post-deploy diff" | Release Verification | | `baseline`, "compare against baseline", "customer org vs our demo" | Baseline Comparison | | Two connector names with no mode word | Ask which mode — the report reads differently in each | All three modes share the same engine (Phases 0–4 below); the mode changes which differences are *findings* and which are *noise*. Never guess the mode: a prod-only Apex class is a red flag in `drift` and expected background in `baseline`. ## Two connectors, one direction Every run names two connectors explicitly: - **Source** — the org who