# Every Salesforce org's configuration and source, one search away. > The SyncOnAI 360 Metadata Repository is an indexed copy of each connected Salesforce org's metadata, including the full source of Apex, Flows and Lightning Web Components. It refreshes incrementally every six hours and on demand, and it powers search, the AI chat, health, the audit and the dependency graph. Salesforce stays the authority. Source: https://synconai360.com/features/metadata-repository ## Key facts - **20+**: Metadata types read, with full source where Salesforce allows - **6h**: Automatic refresh, plus a Refresh button - **Delta**: Later syncs fetch only what changed - **0**: CRM records stored ## Why exploring a Salesforce org is so slow Setup shows one component at a time. Answering a simple question, where a field is used or which classes touch an object, means opening page after page or retrieving metadata into an IDE first. - **Setup is one screen at a time.** There is no single place to search across Flows, Apex, components and configuration together. - **Retrieval needs tooling.** Getting the full source of an org usually means a developer, a project file and a retrieve command. - **Copies go stale.** A one-off export is out of date by the next deploy, and nobody knows how stale. - **AI has nothing to read.** Without the org's metadata in reach, AI assistants answer from general knowledge. ## What does the Salesforce metadata repository hold? Objects, fields, record types, layouts and Lightning pages; Apex classes and triggers with full source; every file of each Lightning Web Component; Flows, validation rules, permission sets and profiles; and email templates, value sets, approval processes, queues, reports, dashboards and sharing rules, along with Apex coverage from the org's last test run. Managed package code that its publisher hides is shown as hidden, never as missing. Your records, accounts, contacts, cases and the rest, are not read into the repository; record queries run live against the org when they are needed. - Configuration and full source together - Package code shown as hidden by its publisher - Records stay in Salesforce ## How does the repository stay current? The first sync of a large org can take several minutes and runs in the background, with progress by stage and counts per metadata type, and an estimate based on the org's own previous syncs. When no estimate can be made honestly, it says so. A running sync can be cancelled. Later syncs fetch only what changed and are much faster. Every org refreshes automatically every six hours, Refresh runs a sync at any time, and components deployed through SyncOnAI 360 are read back straight after a successful deploy so the repository reflects them immediately. - Background syncs with progress by stage - Incremental after the first - Six-hour refresh, or on demand - Deploys read back immediately ## How do you search and explore Salesforce metadata? Search a component by name in the Knowledge Hub and open it to its full analysis, or ask the chat in plain language: which classes reference Opportunity, which Flows run on Case, what writes to a field. Each answer shows the tools it used, so you can see it came from the repository. Open any component in Lineage to see what uses it and what it uses, or compare metadata between two orgs, or between snapshots of the same org, to see how environments differ before a release. - Name search in the Knowledge Hub - Plain-language questions in chat - Compare between orgs or snapshots ## What does the metadata repository power? Everything that needs to know the org works from the same copy: the AI chat and agents, the health score, the audit, the dependency graph and lineage, the builders, the change history and the read-only MCP connection for Claude and Cursor. One copy means every view agrees with every other. Coverage equals what the last sync retrieved, and the platform treats it that way: the absence of a record in the repository is not taken as proof that something does not exist in the org. - One copy behind the AI, health, audit and graph - Read by Claude and Cursor through MCP - Absence never treated as proof ## How the repository is built 1. **Connect.** Authorise a sandbox or production org with Salesforce OAuth. 2. **First sync.** Configuration and source are read in the background with progress by stage. 3. **Explore.** Search, ask the chat, open lineage or compare orgs. 4. **Stay current.** Incremental syncs every six hours, on demand, and after every deploy. ## Exploring an org, with and without a repository | Without | With SyncOnAI 360 | |---|---| | Setup, one screen at a time | One search across the whole org | | Retrieve commands to see source | Full source already synced | | Exports that go stale | Incremental refresh every six hours and after deploys | | Different tools disagreeing | One copy behind every view | | AI with no org to read | AI and MCP clients reading the same copy | ## Who relies on the metadata repository - **Salesforce admins.** Answer where a field is used, which Flows run on an object or what a profile grants in one search, instead of an afternoon in Setup. Plain-language questions in the chat work for admins who do not read Apex. - **Salesforce developers.** Read the full source of Apex, Flows and every Lightning Web Component file without a retrieve command or a project file, and compare a sandbox with production before a release to see exactly what differs. - **Consultancies.** Connect a new client's org on day one and have a searchable copy of everything in it within minutes, so discovery starts from the org itself rather than from interviews and screenshots. - **Your own AI tools.** Claude Desktop, Cursor and other MCP clients read the same copy through a read-only, scoped and logged connection, so their answers are about your org rather than Salesforce in general. ## Frequently asked questions ### What is a Salesforce metadata repository? A searchable copy of an org's metadata: its configuration and the source of its code and automation. SyncOnAI 360 keeps one per connected org and refreshes it automatically. ### Is this a backup of our org? No. It is a working copy for understanding and changing the org. Salesforce remains the authority, records are not included, and it is not a backup or restore product. ### Which metadata types are included? More than twenty, including objects, fields, record types, layouts, Lightning pages, Apex, Lightning Web Components, Flows, validation rules, permission sets, profiles, email templates, value sets, approval processes, queues, reports, dashboards and sharing rules. ### How long does the first sync take? Several minutes for a large org, in the background. Later syncs only fetch what changed and are much faster. ### How often does it refresh? Every six hours automatically, whenever you press Refresh, and immediately for components deployed through SyncOnAI 360. ### Can we see managed package source? Where Salesforce hides it, no. Those components are listed and shown as hidden by their publisher, never as missing. ### Can we compare two orgs? Yes. Metadata can be compared between orgs or between snapshots of the same org. ### Can our own AI tools read the repository? Yes. Claude Desktop, Cursor and other MCP clients can search it through a read-only, scoped and logged connection. ### Is it included on the free plan? Yes. Every plan includes every feature. Free covers one user and one production org with its sandboxes.