PL-400 deep review — September 27, 2026¶
The complete guide and retained March 19 objective snapshot were read. All 88 published October replacement objectives in 15 groups were mapped to guide sections. The PL-400 study-guide URL now contains that future scope; this does not replace the accepted March baseline before the transition. October registration and final delivery dates remain explicit.
Learning improvements¶
Sections 1–6 retain PL-400 preparation; Section 7 expands AB-400 transition work for code apps, APIs and agent integration. Current documentation corrected or sharpened several important boundaries:
- Personal pipelines cannot be extended; advanced requirements need a custom host.
- Connector import format and OAuth support depend on the particular route.
- Attribute presence is different from a changed business value, and database rollback cannot undo a completed external side effect.
- Full synchronization does not replay historical deletions, and elastic Upsert does not raise separate Create/Update events.
- Managed-identity version 2 uses certificate distinguished-name hashes.
- Background operations preserve plug-in timeouts and do not cancel running work.
- Code-app CSP, publication grants, flow types and environment variables have distinct boundaries; agent connector, client-library, MCP and A2A routes differ.
There are four worked examples, ten labs and 44 answered checks. Four synthetic attribute/state-set assertions passed locally. No SDK, platform, Azure, tenant or agent lab was executed.
Blog decision¶
Suyash Kshirsagar's Dataverse Skills: Your Coding Agent Now Speaks Dataverse, published April 1, 2026, was accepted for its tool-orchestration explanation. The public article was read. The guide adds a synthetic two-table review exercise and qualifies setup claims using current preview, privilege, consent and client-allowlisting documentation. No plugin was installed, video played or tenant demonstration executed.
Tiffany Treacy's September 17 Power Platform update was used as a release and developer-documentation discovery channel. Relevant primary pages were followed; its product-level announcements are not blanket GA evidence for every feature.
Evidence and limits¶
ADLC_Docs/operations/2026-09-27-pl-400-deep-review.json preserves prior records,
guide hashes, objective positions/hashes and mappings, source observations,
findings, blog intake and validation. The objective count describes October scope;
the March snapshot and effective-date follow-up remain intact.
The Udemy page was automation-blocked. Public course metadata confirms five days of instructor-led delivery, but the earlier 25-hour Learn-path total is historical. Reachable catalog/practice shells do not demonstrate paid lesson access or an assessment attempt. Preview, host, identity, licensing and deployment behavior must still be tested in a suitable tenant. Research and repair used the same AI context; independent and human review remain pending.
Publication uses the unit-test, repository, strict site-build, generated-site and diff checks. See the Microsoft review tracker for the remaining queue.