microsoft/dataverse-skills
6 skills171 installs816 starsGitHub
Install
npx skills add https://github.com/microsoft/dataverse-skillsSkills in this repo
1Dv ConnectDv-connect (MCP configuration reference) helps solo and small-team builders attach Microsoft Dataverse to the agent they already use—GitHub Copilot or Claude Code—via the Dataverse MCP server. It assumes you have or will obtain a Dataverse environment URL through the dv-connect discovery flow and stores it as DATAVERSE_URL. The skill walks through picking the host tool, assigning the correct MCP client identifiers Microsoft documents for each path, and filling MCP server parameters without unnecessary back-and-forth when env vars and editor context are enough. It is for builders shipping business apps on Dataverse who want Copilot or Claude to query and act on live org data through MCP rather than bespoke REST glue. Pair it with the rest of the dataverse-skills pack for auth and table operations after MCP is reachable.32installs2Dv Metadatadv-metadata is a Dataverse-focused agent skill that teaches how to model alternate keys so bulk upsert paths behave predictably for solo builders wiring CRM or line-of-business data into Power Platform. Without an alternate key, Dataverse cannot match incoming rows to existing records during UpsertMultiple, which silently creates duplicates on every re-run. The skill walks through when to create keys—always on source-system identifiers during schema setup—and how to pick columns per source type: unambiguous PKs from SQLite or SQL Server, proposed candidates from Excel and CSV with explicit user confirmation, and direct questions when no natural key exists. Composite source keys map to composite alternate keys. The excerpt emphasizes doing this before import so migrations stay repeatable for indie integrations and small SaaS backends on Dataverse. It pairs with other dataverse-skills for auth and ingestion rather than replacing full ETL design.31installs3Dv Solutiondv-solution is a Microsoft Dataverse agent skill for solo builders and consultants who must treat customizations as portable solution packages rather than manual clicks in each environment. It documents PAC CLI flows to create solutions, export and unpack artifacts, pack and import into target tenants, and validate deployments—while steering publisher and solution record creation through the Python SDK to dodge urllib encoding and GUID parsing pitfalls called out in SKILL.md. The skill draws hard boundaries: table and form work belongs in dv-metadata, CRUD in dv-data, reads in dv-query, and connectivity in dv-connect. Publisher prefix choice (never default new) is emphasized because it locks naming for tables, columns, and relationships. After import, post-import validation closes the loop so promote-to-prod is evidence-based, not hope-based.31installs4Dv Overviewdv-overview is Microsoft’s router skill for the Dataverse skills pack. Solo builders and small teams touching Dynamics 365, Power Platform, or CRM agents install it so the assistant does not guess which specialist file to open. It owns what no single task skill should: the skill index, shared UX principles, environment confirmation, and pull-to-repo expectations. Before any implementation, it enforces checking initialization—if `.env` or `scripts/auth.py` is absent, the agent must run the dv-connect flow automatically rather than asking permission or inventing its own auth scaffold. Per-request routing is not duplicated here; each downstream skill’s frontmatter triggers decide applicability. Use this whenever the user mentions Dataverse or CRM broadly; then follow the routed skill for queries, metadata, solutions, security, or administration. It keeps multi-phase CRM work coherent from first connect through ongoing schema and ops changes.29installs5Dv Datadv-data is a Microsoft Dataverse agent skill for solo builders and small teams migrating or syncing data from SQLite, SQL Server, Excel, or CSV into Dataverse without breaking foreign keys or duplicating rows on reruns. It prescribes a strict pipeline: create tables with prefixed source ID columns using dv-metadata, register alternate keys on those columns, wire lookup relationships, then import in dependency order via UpsertItem keyed on business IDs rather than GUIDs. That design makes partial failures and replays idempotent—a common pain when agents blindly insert parent and child rows. The skill distinguishes database sources, where the agent can infer primary keys from schema, from spreadsheet sources, where it must inspect uniqueness and ask the user to confirm the business key before creating an alternate key. It targets indie SaaS on Power Platform and consultants wiring line-of-business data into model-driven apps. Prerequisites assume the Dataverse skills repo scripts and auth helper are in the workspace.24installs6Dv Querydv-query is a Microsoft dataverse-skills agent skill for solo builders and consultants who need to pull Dynamics 365 / Dataverse data into Python for exploration, reporting, or agent-assisted analysis. It documents notebook-first setup with InteractiveBrowserCredential, constructing a DataverseClient, and loading entities such as accounts into pandas via dataframe.get with selective columns. For SDK b8 and later, it introduces QueryBuilder as the fluent query path; earlier versions must use records.get instead of introspecting undocumented APIs. The skill explicitly separates notebook auth from script auth when scripts/auth.py exists in sibling skills. Builders use it during integrations work when CRM data must join internal analytics, prototypes, or automation. It is integration documentation packaged as a skill, not a hosted query service.24installs