Add blog post: Dataverse searchQuery as a discovery agent in Copilot Studio - #1
Open
KarimaKT wants to merge 73 commits into
Open
Add blog post: Dataverse searchQuery as a discovery agent in Copilot Studio#1KarimaKT wants to merge 73 commits into
KarimaKT wants to merge 73 commits into
Conversation
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Covers six ways to access Copilot Studio conversation transcripts, what diagnostic data is available, required roles, and AI-powered transcript analysis. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add "Understanding the data model" section explaining records, sessions, and conversations with mermaid timeline diagram. Add "Where to find what" section with persona-based tool recommendations and mermaid flowchart. Update App Insights section with session outcomes clarification and Analytics Template Workbook link. Update TL;DR with new content pointers.
…d 3) - Remove all em dashes, replace with hyphens (text) and double dashes (code blocks) - Fact-check App Insights section: add designMode filtering, user_Id limitation, three logging toggles, retention clarification - Add push/pull distinction to Dataverse vs App Insights table - Clarify session timeout is analytics-only boundary, does not reset conversation history or LLM context - Consolidate dev environment warning and PII/security guidance into single callout - Add summary table to six ways section with numbered expandable items and blue click-to-expand cues - Collapse ready-to-use prompt into expandable section - Fix MCS Agent Analyser: deterministic not AI-powered, prominent link - Rewrite draft note on line 191 - Add (now Microsoft Entra ID) on first Azure AD mention - Reorder What to do next: start with eval-driven development - Add engaging ending with call to action for comments
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…nd 5) - Add companion technical reference post with data model, access methods, and code examples - Add distinct header image for technical reference post - Fix "message credit mechanism" → "Copilot Credits" with updated URL - Fix "extended thinking" → "deep reasoning" (Microsoft terminology) - Define CSAT, KQL, and MSAL on first use with hyperlinks - Add anchor links for persona sections in TL;DR - Add comment for unused json import in code example
- Fix KQL code fence rendering (kql → sql) - Fix DesignMode casing mismatch in tip vs KQL script - Remove duplicate conversationId extraction in KQL - Update Synapse Link references to Fabric link - Remove hallucinated Agent Insights Hub references - Fix chain-of-thought reasoning claim - Add data model overview image to both posts - Wrap KQL trace query in collapsible details block - Add cross-references between posts (triage ↔ KQL) - Differentiate tags between posts - Minor readability and formatting improvements
- Restructure Scenario 1: snapshot instructions before example, real example as payoff - Reframe Scenario 2 around persona (ops/support in App Insights), not timing - Clarify what App Insights does/doesn't show (tool invocations as TopicStart, no KS details) - Reframe Scenario 3: highlight built-in analytics (themes, answer quality, sentiment, savings) - Reposition Copilot Studio Kit as custom analytics middle ground, not "deeper analytics" - Add SharePoint sensitive data exclusion to "What's not in the transcript" - Reference community tools (MCS Agent Analyser + Vineet's agentic tooling) in context - Remove redundant sections: What to do next, Beyond transcripts, Quick reference, Roles - Remove technical reference post links (deferred to follow-up) - Tighten closing to frame post as starting point with follow-ups based on interest Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…al review Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This reverts commit a3dda97.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
microsoft#239) * Add blog post: Dataverse searchQuery as a discovery agent in Copilot Studio - New post: unauthenticated fuzzy search over structured Dataverse data - Covers searchQuery + List Rows two-tool pattern, facets, caching, child agents - Add author entry for KarimaKT - Include 9 screenshots with captions and shadow styling * Update Dataverse Search blog post and add images * Fix PowerFx validation syntax in guardrail code block * Update blog post to v5 - final version * Update blog post content --------- Co-authored-by: KarimaKT <kkanjitajdin@microsoft.com>
…ry blog post - Rewrote intro to lead with unauthenticated + structured data problem - Restructured for progressive disclosure: basic one-tool pattern first, enhancements optional - Split full YAML dump into basic config + progressive enhancement snippets - Moved List Rows from basic to Making It Smarter (searchQuery returns full multiline content) - Added OData filtering, synonym enrichment, pagination as separate opt-in sections - Replaced emoji callouts with Chirpy prompt boxes throughout - Added links to MS docs for every setup step, auth config, and connector reference - Added internal links to Dataverse MCP post and custom knowledge source post - New header image, new tool config screenshots - Removed glossary, Go Deeper, Further Reading sections - Cut Enterprise Patterns, Topic Caching, PowerFx guardrail (scope reduction) - Fixed broken image syntax, unused images, collapsible rendering - Made all YAML blocks collapsible with proper HTML rendering Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New video demo for the plugin
Replace 14-category scoring system with priority-based review: 1. Narrative and structure (problem framing, progressive disclosure, scope) 2. Technical accuracy (verifiable claims, doc links, internal links) 3. Reader experience (image placement, collapsibles, Chirpy conventions) 4. Front matter and metadata (tags for Further Reading overlap) 5. Scope and length (one thing well, flag tangential sections) Output is now ranked issues with quoted text, not percentage scores. Updated: .github/instructions/posts.instructions.md (GitHub Copilot) Updated: .claude/skills/review-post/skill.md (Claude Code skill) Updated: CLAUDE.md (reference to new approach) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Writers can copy the skill to their local Claude Code setup and run /review-post to get editorial reviews of blog post PRs. Includes README with install instructions and explanation of why this complements (not replaces) GitHub Copilot's automated review. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The skill is already tracked at .claude/skills/review-post/SKILL.md and auto-discovered by Claude Code. Removed duplicate skill.md from tools/. Updated README with correct usage instructions and two-layer review strategy (GitHub Copilot for linting, Claude Code for editorial). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
… Copilot CLI Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- /local-server: starts Jekyll dev server and opens the post - /submit-pr: commits changes, creates branch, opens PR with checklist Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- mcscatblog-review-post - mcscatblog-local-server - mcscatblog-submit-pr Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds bundle check / bundle install step before starting the server, so writers don't need to install Ruby dependencies manually. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Points to tools/setup/mac/install.sh and tools/setup/win/install.ps1 for first-time setup (Ruby, Bundler, bundle install). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Creates branch, post file with front matter, and asset folder. Sets published: false by default. Guides writer through the full lifecycle using the other skills. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New blog post covering the dynamic MCP routing sample from CopilotStudioSamples. Walks through catalog-driven endpoint discovery, x-ms-dynamic-values dropdowns, and C# URL rewriting for routing a single connector to multiple MCP endpoints. Also fixes mcscatblog-new-post skill to not set published: false (blocks local preview on feature branches). Co-authored-by: adilei <adileibowiz@microsoft.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Update lib * The Orchestrator’s Secret: Chain-of-Thought & Transcript on Demand Topics are where the secret switches are hidden in Copilot Studio. Two under-the-radar tricks: surfacing the orchestrator's chain of thought as it works, and pulling the full conversation transcript into a variable on demand. * Update 2026-03-13-power-of-topics-copilot-studio.md * updated images * Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> * Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> * Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> * Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> * updated screenshots * Update 2026-03-13-power-of-topics-copilot-studio.md * Review edits: clarity and technical accuracy improvements - Explain core mechanism (variable descriptions as orchestrator instructions) - Clarify CoT is elicited reasoning, not raw reasoning tokens - Add transcript accuracy caveat (best-effort, not compliance-grade) - Explain Recognize intent node with prompt-tip callout - Add recursion guard caveat - Fix broken image reference (conv-history-topic.png → cot-topic.png) - Condense security/privacy checklist - Minor clarity fixes (end user vs user, / picker, inside the dialog) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * updating screenshots and adding missing one * updated screenshots * Should prompt user should be unchecked Make sure to uncheck the box "Should prompt user" in the "Additional settings" section at the bottom --------- Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> Co-authored-by: adilei <adileibowiz@microsoft.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This article covers 6 production-grade patterns for Copilot Studio agents in Microsoft Teams, with deep focus on self-service diagnostic Adaptive Cards that reduce support burden. Key topics: - OnInstallationUpdate for handling reinstalls - Session expiration with OnInactivity triggers - Context variables that work in M365 Copilot - Enhanced reset conversation flow - Diagnostic cards with expandable debug menu (hero feature) - Suggested prompts configuration Target audience: Makers building production Teams agents Word count: ~4,100 words Screenshots: 18 embedded images Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
… agents Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…aming - Remove "Color Psychology for Environments" section - Standardize environment references to uppercase (DEV, TEST, PROD) - Fix image numbering: repo images were shifted by 3 positions vs Word doc - image21 now shows manifest.json (was docx image24) - image22 now shows Copilot Studio sharing (was docx image25) - image24 now shows Edit availability (was docx image26) - image25 now shows Setup policy (was docx image27) - image26 now shows Assign policy (was docx image28) - image27 now shows M365 availability (was docx image29) - All images are uncompressed originals from the Word doc Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Remove first environment table (covered by deployment path comparison) - Remove decision matrix (third table saying the same thing) - Remove "Deploying DEV and TEST Environments" (repeats manifest section) - Remove "Manifest Customization Deep Dive" (repeats Step 3) - Trim "Why This Matters" to two sentences - Remove Troubleshooting section - Update checklist: remove diagnostic cards, add setup policy conflict check (policies are not cumulative, highest-ranked group assignment wins) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix "Controlled distribution" to clarify Copilot Studio sharing vs Teams app distribution are separate concerns - Remove "Governance requirements" bullet (Option 2 has approval) - Clarify both Option 1 methods are sideloading, mention sharing via link - Update Option 2: admin can auto-install/pin after approval (same as Option 3), update comparison table accordingly - Add link to full M365 app manifest schema reference Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add tip note about clearing history reducing token limit errors - Fix On Error section: add missing images (image19-21) - Fix suggested prompts section: use correct images (image22-23) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Replace image18 with correct Start Over expanded flow screenshot - Remove redundant image from The Result section (intro GIF already shows final product) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add callout in M365 Copilot section: channels behave differently, test both independently - Add M365 Copilot validation step to production checklist Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Explains why this startBehavior is needed: Teams conversations persist longer, so without it users can get stuck on older published versions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…eams-agent-patterns Add Teams agent patterns post with solution download
…eams-deployment Update Teams deployment post date to 2026-04-07
- Fix relative image path for GIF (../assets → /assets) that broke htmlproofer - Fix concatenated YAML keys (startBehavior and beginDialog on same line) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…rns-build Fix broken image path and malformed YAML in teams agent patterns post
The zip download link used an absolute path without the site baseurl (/mcscatblog), causing htmlproofer to fail. Chirpy transforms image paths automatically but not href links. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Fix download link missing baseurl prefix
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…-image Replace deployment post header image
New post covering 5 Dataverse retrieval methods for Copilot Studio agents: Knowledge, MCP Server, List Rows, Search Query, and Prompt Tool. Includes decision flowchart, setup steps, and trade-off comparison.
…eractive widget
- Rewrote scenario section: replaced 6-column markdown table with real CSV-backed
Dataverse screenshot and prose describing the 13-column dataset
- Updated all logical names from placeholder cr_* to real crc57_* names matching
the searchquery blog (crc57_facility, crc57_district, crc57_facilitytype, etc.)
- Updated facility types and districts to match actual CSV data
- Activated all 7 screenshots with accurate captions
- Rewrote Prompt Tool section: shifted focus from aggregation to semantic reasoning
('Where can I play basketball?') as the core differentiator
- Updated Decision Matrix with nuanced accuracy across all methods
- Rewrote flowchart decision tree for clearer method selection
- Added interactive React widget (71KB, Preact + Tailwind) showing retrieval shape
per method with color-coded table highlighting and toggle between table/details
- Added react-widget build skill for future widget embedding
- Updated .gitignore for widget build artifacts
- Fix broken header image (placeholder from existing asset) - Fix front matter: tags typo (period→comma), replace generic tags, fix description typo - Shorten mermaid decision flow node text for readability - Fix typos: semandic→semantic, agentès→agent's, missing prompt-info class - Wrap long code blocks in collapsible <details> sections - Expand Search Query "hit the wall" section with concrete example - Reference unused images (TableImported01, 01tabledatafacilities) in Scenario section Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Reword Q1 from "Are queries specific and filterable?" to "Can user requests map to precise queries?" to clarify that List Rows works for complex NL requests mappable to OData filters. Pull Prompt Tool out of the fuzzy/discovery branch into its own top-level "Need reasoning over content" path from Q1, since LLM reasoning over tabular data is orthogonal to fuzzy search.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…Studio