A public toolkit of reusable skills, tools, and scripts for SharePoint engineering, modernization, and knowledge transformation:
- SharePoint Site Migration: End-to-end site discovery, dependency wave sequencing, link remediation, and content migration.
- SharePoint Object Creation: Automated, declarative provisioning of lists, document libraries, views, site columns, content types, and permissions.
- Publishing & Modifying Content: Modernizing classic ASPX pages, authoring/updating modern SharePoint pages, and publishing Markdown/assets.
- Creating & Publishing Copilot Agents & Skills: Authoring, validating, deploying, and managing SharePoint Copilot agents (
.agentpackages) and native Copilot Studio skills in SharePoint. - Content Maintenance, Continuous Improvement & Document Decomposition: Breaking huge documents and legacy manuals (Word/PDF) into modular, maintainable subpages in SharePoint, continuously improved and supported by Copilot agents and skills (Content + Template + Renderer = Published Output).
The SharePoint Knowledge Workbench provides a modular ecosystem of unified plugin with 11 domain-focused skills and 9 sub-agents and associated agent skills/scripts designed to bridge SharePoint site modernization, automated object provisioning, and intelligent knowledge management.
Rather than maintaining monolithic Word/PDF manuals or writing one-off migration scripts, the workbench enables teams to decompose massive documents into structured subpages maintainable directly in SharePoint, while providing enterprise-grade tooling to plan and execute SharePoint migrations, provision schema and objects, deploy Copilot Studio agents/skills, and maintain high-fidelity SharePoint sites.
See INSTALL.md for installation and consumer integration options, or browse docs/use-cases/ for detailed guides across all functional areas.
+-----------------------------+
| Legacy Intake (.docx) |
+-----------------------------+
|
v plugins/ (4 domain plugins)
+-----------------------------+
| Analysis & Cleanup Plan |
+-----------------------------+
|
v TDD Validation
+-----------------------------+
| Structured Markdown Core |
| (Single Source of Truth) |
+-----------------------------+
|
+------------------------+------------------------+
| |
v v
+----------------------------------+ +----------------------------------+
| Human-Facing Renderer | | Agent-Optimized Renderer |
+----------------------------------+ +----------------------------------+
| |
v v
+----------------------------------+ +----------------------------------+
| Target A: Formatted Page / ASPX | | Target B: Agent Digest / Index |
| - Rich layout, images, CSS | | - High semantic density |
| - Designed for visual reading | | - Embedded source link pointers |
+----------------------------------+ +----------------------------------+
The assessing-sharepoint-sites and migrating-sharepoint-sites skills support recursive CSV file inventories, selected bulk downloads, and offline Python link inventories for static HTML/HTM/ASPX and Office documents. An Online collector exports modern-page stored fields for the same Python analysis.
# User-run file inventory: choose the target and destination folder.
pwsh -File plugins/sharepoint-workbench/.apm/skills/assessing-sharepoint-sites/scripts/collect-sharepoint-content-inventory.ps1 -SiteUrl https://sharepoint.example.org/ -OutputDir ./inventory
# Offline Python analysis after selected downloads.
python plugins/sharepoint-workbench/.apm/skills/modernizing-sharepoint-pages/scripts/link-remediation/export_link_inventory.py --manifest-csv ./downloads/downloads.csv --output-dir ./linksThe collector writes files.csv, including source site/subsite, container/library, path, filename, extension and metadata; it also writes library/error CSVs and a status manifest. Online hostnames select interactive browser sign-in; on-prem prompts for credentials. No file contents are downloaded by the inventory collector. Link extraction retains original source URLs and reports failed/unsupported sources.
See the bulk content/link workflow for download, modern-field export, conversion, rewrite, publishing and validation stages.
Skip this step if you only need the document-conversion pipeline (no live SharePoint tenant access). Any use case that connects to a real tenant (setting-up-sharepoint-workbench's live network/auth checks, publishing-sharepoint-pages, live discovery/migration work) needs an Entra ID app registration your tenant administrator creates — request it before starting.
Two distinct registration types exist for two distinct purposes — an unattended, App-Only/certificate registration for scheduled jobs (no signed-in user, no MFA), and an interactive/delegated registration for human-operator provisioning and migration work (requires an M365 E3/E5 licence, browser/device-code sign-in). Both use Sites.Selected (Graph and SharePoint, Application permission type) — not the broader AllSites.* grants — which grants zero site access by itself; a tenant administrator must separately grant your specific sites via Grant-PnPAzureADAppSitePermission after admin consent.
The setting-up-sharepoint-workbench skill walks through both types end to end via mode app-registration. See plugins/sharepoint-workbench/.apm/skills/setting-up-sharepoint-workbench/references/effective-permissions-matrix.md for the full permission model and empirical findings.
Once approved, record the Application (client) ID and Directory (tenant) ID — both are needed by init-connection.
Install the workbench using APM (Agent Package Manager) across your active agent harnesses:
# Install globally for Claude, Copilot, Codex, and Antigravity
apm install @sharepoint/sharepoint-workbench --global --target claude,copilot,codex,antigravity
# Or install locally in your active repository
apm install @sharepoint/sharepoint-workbench --target claude,copilot,codex,antigravityAlternatively, install via native Claude Code marketplace:
/plugin marketplace add richfrem/sharepoint-knowledge-workbench
/plugin install sharepoint-workbench@sharepoint-knowledge-workbench
setting-up-sharepoint-workbench provides 5 focused modes:
app-registration— Generates fillable service requests and approval guides for tenant administrators.init-connection— Writes the root, git-ignoredconfig.psd1(site URL, tenant ID, client ID, auth mode).workflow-intake— Interactive intake wizard; configures document-workflow profiles and publication targets.validate-connection— Validates connection parameters, executes live tenant auth probes, and tests permission boundaries.resolve-paths— Resolves document IDs and execution profiles into concrete tool parameters.
- start-here.md — Primary Quick-Start & Status Guide. Read this first for active state, phase gates, and workflow guidelines.
- docs/use-cases/ — All 12 use cases this repository serves (11 built, 1 target-state vision), one overview doc each. Start here if you're looking for "how do I do X" rather than architecture/status.
- architecture.md — System structure, directory maps, and technical component flow.
- CLAUDE.md / AGENTS.md — Agent coding conventions, TDD rules, and project guidelines.
- DEPENDENCIES.md — Running log of external CLI requirements, including Pandoc and LibreOffice.
- Master Initiative Plan and Traceability Matrix — Master plan, workstreams, phases, and traceability.
- Vision Overview — Vision overview and change-control rules.
- Research Index and Decision Guides — Comprehensive research navigation.
- Dual-Target Rendering Model — Human-facing versus agent-optimized output.
- AgentAssets Skill-Creation Field Note — Empirical Phase 3.0 SharePoint
AgentAssetsdiscovery findings. - Skill Runtime Decision Guide — Decision framework for SharePoint-native versus GitHub-repository skills.
The repository maintains formal Mermaid architecture diagrams in docs/diagrams/:
- Overall System Architecture — End-to-end view from
.docxintake to structured storage and multi-runtime agent delivery. - Phase 1 Conversion Pipeline — Stage breakdown (
analyze→confirm→convert→render). - Document Analysis & Plan Confirmation — Interactive human-in-the-loop plan confirmation.
- Structured Content Creation — Pandoc AST cleanup, media extraction, structural anchor resolution, and packaging.
- Render & Multi-Target Generation — Multipage Markdown and dual-target rendering engine.
- Validation & Atomic Promotion — Content-loss prevention, schema validation, and atomic staging promotion.
The repository ships 1 unified domain plugin (@sharepoint/sharepoint-workbench) under plugins/sharepoint-workbench/, comprising 11 domain skills and 9 specialized subagents:
setting-up-sharepoint-workbench— Access, connection configuration, project intake profiles, path resolution, and live tenant verification.assessing-sharepoint-sites— Read-only site inventory, schema drift analysis, calculated/choice column extraction, and assessment reporting.converting-sharepoint-documents— Document decomposition: extraction, structure analysis, package assembly, dual-target rendering, and editorial review.managing-sharepoint-schema— Declarative schema planning and reconciliation (site columns, content types, schema diffing, and scaffolding).managing-sharepoint-lists-and-libraries— Provision and manage SharePoint lists, document libraries, list views, and calendar lists.managing-sharepoint-navigation— Declarative navigation management, Excel/CSV sync, structure extraction, and validation.publishing-sharepoint-pages— Modern page publishing, static ASPX asset upload, Markdown publication, dynamic ASPX page migration, and state verification.modernizing-sharepoint-pages— Classic-to-modern page transformation, web part modernization, page previews, and design-system compliance.migrating-sharepoint-sites— End-to-end migration execution: dependency waves, list content migration, link remediation, and wave scaffolding.developing-spfx-solutions— SPFx solution scaffolding, React/master-detail web parts, extensions, packaging, and App Catalog deployment.managing-sharepoint-copilot-agents-and-skills— Author, deploy, verify, back up, and restore Copilot Studio agents and native SharePoint skills.
sharepoint-discovery-agent,sharepoint-deployment-planning-agent,sharepoint-deployment-sequencing-agent,sharepoint-content-migration-sequencing-agent,sharepoint-modernization-agent,sharepoint-webpart-modernization-analysis-agent,sharepoint-link-agent,sharepoint-link-remediation-analysis-agent,sharepoint-validation-agent.
| Use case | Overview | Skill Hub | What it does |
|---|---|---|---|
| Document conversion & decomposition | overview | converting-sharepoint-documents |
Extract, analyze, assemble and render Word manuals into governed Markdown / SharePoint pages. |
| Site assessment & schema comparison | overview | assessing-sharepoint-sites |
Read-only analysis of classic SharePoint inventories, schema drift, calculated columns, and permissions. |
| Schema reconciliation | overview | managing-sharepoint-schema |
Declarative planning and reconciliation of site columns, content types, and schema exports. |
| Tenant provisioning | overview | managing-sharepoint-lists-and-libraries |
Real PnP executors for lists, libraries, columns, content types, views, and items. |
| Tenant publication | overview | publishing-sharepoint-pages |
Markdown and page publication plans, dynamic ASPX migration, static ASPX asset uploads. |
| Classic-to-modern page conversion | overview | modernizing-sharepoint-pages |
Classic page analysis, modernization planning, single/library conversion, validation and reports. |
| Link remediation | overview | migrating-sharepoint-sites |
Extraction, URL updating and validation of page/document links and rich-text image links. |
| Content migration | overview | migrating-sharepoint-sites |
Item-level list content migration with two-pass lookup-ID re-linking. |
| Migration wave planning | overview | migrating-sharepoint-sites |
Dependency-graph analysis, migration wave planning and wave script scaffolding. |
| SPFx web part development | (no overview doc) | developing-spfx-solutions |
Web part scaffolding, packaging and App Catalog / direct-publication delivery. |
| Workbench setup | overview | setting-up-sharepoint-workbench |
Connection config, document-workflow and publication-profile setup shared by the other use cases. |
| Agent & native-skill lifecycle | overview | managing-sharepoint-copilot-agents-and-skills |
Create/update/deploy/verify/undeploy/backup/restore for SharePoint Copilot agents and native skills. |
The unified package is installable via APM (apm install @sharepoint/sharepoint-workbench). Most write-capable modules across this set share one safety contract: planning is pure (no I/O), apply is dry-run by default, and a real tenant write requires both an explicitly injected executor and a plan-derived confirmation token. For full technical details and transition records, see docs/plans/master_migration_inventory.csv.
Beyond the use cases above (all real, working, tested), the repository has a documented but
not yet implemented target state: moving knowledge management itself — not just the initial
conversion — to an AI-assisted, agent-enabled pipeline. Natural-language library setup, AI-assisted
metadata/classification/cleanup, governed review and approval, SharePoint knowledge agents managed
as a governed product, and continuous knowledge-health monitoring. See
docs/use-cases/ai-assisted-knowledge-management-pipeline.md
for what's real today versus what's still proposed, and
docs/vision/ai-assisted-sharepoint-knowledge-workbench-governance-vision.md
for the full capability vision. Nothing in this section is authorized backlog — it exists to
document direction, not current capability.