Reusable Microsoft administration scripts, their tests, and the conditions under
which they have actually been checked. Evolved from m365management; existing
script paths remain available.
Start with the task, not with collecting more code. Search what exists, read its evidence and limitations, then reuse, improve, or generate what the task needs. Afterwards, preserve the generalized improvement and the observed result.
Requires Python 3.10+; no third-party packages, credentials, or network access. These commands only read metadata and files. They never execute administration code.
python3 tools/toolkit.py search "exchange"
python3 tools/toolkit.py search "棚卸し"
python3 tools/toolkit.py search "azure"
python3 tools/toolkit.py validatecatalog.json lists maintained local scripts. sources.json lists external
references, including Sam Erde, PnP, Azure samples and the existing AzureManagement
repository. Registration is not adoption or endorsement of every script.
| Area | Existing code / next-use references |
|---|---|
| Identity | Active Directory, Entra ID |
| Microsoft 365 | Inventory, Exchange, Teams |
| Azure | AzureManagement and Azure official samples in sources.json |
| Endpoints, security, other Microsoft products | Add the first actual use case; no empty category scaffolding needed |
PowerShell, CLI scripts, Python, KQL and infrastructure templates can all be catalogued. Existing PowerShell scripts must be registered; other file types may be registered using the same metadata format.
All inherited scripts currently start as legacy-unverified. This does not say that they have never worked; it says revision-bound execution evidence has not been recorded here. Existing documentation is not proof of tested permissions or complete result collection. See known limitations.
legacy-unverified: inherited code without current execution evidence.experimental: new or changed code awaiting field validation.verified: a current passed lab/production record exists for all declared artifacts; the claim applies only to that record's conditions.deprecated: read the runbook for the replacement.
An offline test does not establish tenant validation. validate checks record
consistency, not the truth of a recorded claim or the correctness of a script.
Read the contribution workflow. It covers reusable changes, source attribution, evidence records, and upstream updates. A concrete first AD review record demonstrates scoped, revision-bound evidence without claiming live-domain validation.
python3 tools/toolkit.py evidence-template find-oldusers --scope offline
python3 -m unittest discover -s tests -vRun administration code only after checking the target service's authentication, permissions and existing change-approval rules. Cloud CLI, Graph, Exchange, Teams and on-premises AD contexts are independent. Keep tenant configuration, customer outputs and credentials in their existing private stores.