Teams App Developer

MicrosoftOfficialPublished by Microsoft

Skill
Tool integration
Software development
MIT
Medium risk

From the author

Builds, tests, and deploys Microsoft 365 apps and agents for Teams and Copilot. Includes sub-skills for project creation, local testing, cloud deployment, troubleshooting, and Slack-to-Teams migration. USE FOR: Teams agent, bot, tab, message extension, Declarative Agents, Custom Engine Agents, local testing, Agents Playground, Azure resource provision, remote deployment, Slack to Teams migration, cross-platform bot development, Block Kit to Adaptive Cards conversion. DO NOT USE FOR: general web development, non-bot/non-Teams projects.

— the description by Microsoft

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

The teams-app-developer skill provides a structured framework for managing the lifecycle of Microsoft 365 agents and Teams applications. It leverages the ATK CLI to handle project scaffolding, local testing via the Agents Playground, cloud provisioning on Azure, and troubleshooting. The skill includes specialized guidance for migrating Slack bots to Teams, including converting Block Kit UI to Adaptive Cards, and provides access to an extensive library of micro-expert documentation for cross-platform development.

What it does

Scaffolds new Microsoft 365 agents and Teams applications from templates.
Facilitates local testing using the Agents Playground or Teams-specific environments.
Automates Azure resource provisioning and cloud deployment workflows.
Provides diagnostic procedures for common authentication and configuration errors.
Supports migration of Slack bots to the Teams platform.
Manages project configuration and environment variables for ATK-based projects.

When to use it

When starting a new Teams app, bot, or Declarative Agent project.
When you need to test an agent locally before deploying to the cloud.
When migrating existing Slack bots to the Microsoft Teams ecosystem.
When troubleshooting deployment or authentication issues in a Teams project.
When you need to provision Azure infrastructure for a Teams application.

Advantages

Streamlines the development lifecycle with standardized CLI commands.
Reduces development time by offering a local testing playground.
Provides expert-level guidance for complex cross-platform migrations.
Ensures consistent configuration management across development environments.
Includes robust error-handling protocols for common platform-specific pitfalls.

Usage examples

“Create a new Teams bot project using the default template.”: The agent will initialize a new project folder, configure the necessary YAML files, and guide you through the initial setup using the ATK CLI.
“How do I migrate my Slack bot to Microsoft Teams?”: The agent will initiate the slack-to-teams workflow, providing access to cross-platform bridging experts and guidance on converting UI components.
“My Teams bot is returning a 401 error, how do I fix it?”: The agent will trigger the troubleshooting workflow, checking your environment variables, tunnel status, and authentication configuration.

Install

Download the skill as a ZIP:
https://autogineer.com/agent-skills/microsoft/teams-app-developer/download

Unzip it into the cross-client skills folder of your project:

mkdir -p .agents/skills && unzip teams-app-developer.zip -d .agents/skills

Documentation by Microsoft

Microsoft 365 Agents Toolkit Skill

Build Microsoft 365 agents and Teams apps using the ATK CLI.

AI Behavior Guidelines

Testing Strategy: Recommend Agents Playground first (faster, no M365 needed). Use Teams workflow only if user explicitly requests it.
Environment Variables: NEVER hardcode secrets or make up placeholder values. Always ask users for real values.
Error Handling: Read error messages carefully. Check env/.env.local, .localConfigs, and atk auth list. Common pitfalls:
- AADSTS7000229 → aadApp/create missing generateServicePrincipal: true in YAML — add it and re-provision
- Missing TENANT_ID in .localConfigs → SDK uses wrong token authority → 401 from Bot Connector
- 401 persists after auth fix → devtunnel URL may be blacklisted — create a fresh tunnel
- See troubleshoot/troubleshoot.md for full diagnostic steps
Long-Running Commands — WAIT for completion:
- atk new, atk provision, atk deploy can take several minutes
- Always wait for completion before running the next step (timeout 120000ms+)
Local Service Startup — Hangs terminal (expected):
- npm run dev, npm start, python app.py, devtunnel host, etc. will hang — the process keeps running indefinitely
- ALWAYS run as a background process (isBackground=true) — NEVER use isBackground=false for these commands
- Do NOT wait for it to "finish" — verify startup by checking output for "listening on port" or tunnel URL
- If errors appear, read logs, diagnose, fix, restart
- Use a NEW terminal to launch Agents Playground or open Teams sideloading URL
Monitor App Logs: Periodically check background terminal output for runtime errors. If the app crashes, read the error, fix the root cause, and restart.
Telemetry Tagging: Before running any atk CLI commands, set the session environment variable so all CLI invocations are tagged as skill-initiated:
```
export ATK_CLI_SKILL=true
```
Run this once at the start of the session. All subsequent atk commands in the same terminal will inherit it.

ATK CLI Setup

atk --version  # Must be > 1.1.5-beta

If ATK is not found or version is too old:

npm i -g @microsoft/m365agentstoolkit-cli@beta

CLI Global Options

Option	Meaning	Recommendation
`-i`	Interactive mode	Always use `-i false` in automation to avoid hanging
`-f`	Project folder	Default to be current directory, used when specifying a custom folder. When scaffolding a new project, this is the parent folder where the project folder will be created under.
`-h`	Command help	Use `atk <command> -h` for quick syntax checks

Sub-Skills

Sub-Skill	When to Use	Reference
create-project	Scaffold new project from template, choose template, `atk new`	create-project/create-project.md
test-playground	Test locally with Agents Playground, `agentsplayground`, quick testing	test-playground/test-playground.md
test-teams	Run on Teams, devtunnel, sideload, Teams testing, test in Copilot	test-teams/test-teams.md
provision-deploy	Provision Azure resources, deploy to cloud, `atk provision`, `atk deploy`	provision-deploy/provision-deploy.md
troubleshoot	Fix errors, 401, port conflicts, YAML errors, stale bots	troubleshoot/troubleshoot.md
slack-to-teams	Migrate Slack bot to Teams, cross-platform bridging, Block Kit to Adaptive Cards	slack-to-teams/SKILL.md

MANDATORY: Before executing any workflow, read the corresponding sub-skill document.

Shared References

manifest-and-yaml.md — Project files, YAML config, env vars, .localConfigs flow
commands.md — ATK CLI commands: package, validate, share, collaborate
templates.md — Complete template catalog with language support
experts/ — 100+ micro-expert files: Teams SDK, Slack SDK, cross-platform bridging, deploy, AI models, security, language conversion
docs/ — Platform comparison guides: UI, messaging, identity, infrastructure, feature gaps

Workflow Chains

Match user intent to the smallest valid workflow.

User Intent	Workflow (read in order)
Build new app from scratch	create-project → test-playground
Test existing project locally	test-playground (recommended) or test-teams
Deploy to Azure	provision-deploy
Fix broken bot	troubleshoot → re-test
Migrate Slack bot to Teams	slack-to-teams

MANDATORY: Before executing any slack-to-teams workflow, read slack-to-teams/SKILL.md first. The sub-skill contains a routed expert system with 100+ micro-expert files for cross-platform bot development.

ATK Project Context Resolution

Resolve config values only when missing. If a value is already known in the session, reuse it.

Step 1: Detect ATK Project

If m365agentstoolkit*.yml exists in the current folder, treat it as an ATK project and parse configuration.

Step 2: Resolve Common Configuration

Resolve variables referenced in m365agentstoolkit*.yml. Common variables: AZURE_OPENAI_API_KEY AZURE_OPENAI_ENDPOINT AZURE_OPENAI_DEPLOYMENT_NAME

Step 3: Collect Missing Values

If required values are missing, ask the user for only the missing ones.

Refer to manifest-and-yaml.md for full config-file details.

Files

ATTRIBUTION.mdAdded by AutoGineer906 B
LICENSELicence file added by AutoGineer1.1 KiB
SKILL.md7.6 KiB
create-project/create-project.md8.1 KiB
docs/README.md2.8 KiB
docs/advanced-features.md7.0 KiB
docs/feature-gaps.md46.3 KiB
docs/files-and-links.md4.2 KiB
docs/identity-and-auth.md4.3 KiB
docs/infrastructure.md6.7 KiB
docs/interactive-responses.md4.4 KiB
docs/messaging-and-commands.md4.5 KiB
docs/middleware-and-handlers.md4.4 KiB
docs/ui-components.md6.9 KiB
docs/workflows.md14.4 KiB
experts/README.md8.6 KiB
experts/_expert-ts.md2.2 KiB
experts/analyzer.md14.0 KiB
experts/bridge/app-distribution-packaging-ts.md14.2 KiB
experts/bridge/channel-ops-graph-ts.md15.4 KiB
experts/bridge/commands-slash-text-ts.md18.5 KiB
experts/bridge/cross-platform-advisor-ts.md36.4 KiB
experts/bridge/cross-platform-architecture-ts.md8.3 KiB
experts/bridge/events-activities-ts.md22.4 KiB
experts/bridge/files-upload-download-ts.md17.2 KiB
experts/bridge/identity-oauth-bridge-ts.md19.8 KiB
experts/bridge/index.md13.0 KiB
experts/bridge/infra-compute-ts.md14.6 KiB
experts/bridge/infra-observability-ts.md14.0 KiB
experts/bridge/infra-secrets-config-ts.md13.2 KiB
experts/bridge/infra-storage-ts.md14.3 KiB
experts/bridge/interactive-responses-ts.md19.6 KiB
experts/bridge/link-unfurl-preview-ts.md17.1 KiB
experts/bridge/middleware-handlers-ts.md15.4 KiB
experts/bridge/python-cross-platform.md9.3 KiB
experts/bridge/rate-limiting-resilience-ts.md17.6 KiB
experts/bridge/rest-only-integration-ts.md8.4 KiB
experts/bridge/scheduling-deferred-send-ts.md17.8 KiB
experts/bridge/shortcuts-extensions-ts.md17.5 KiB
experts/bridge/transport-socketmode-https-ts.md14.3 KiB
experts/bridge/ui-app-home-personal-tab-ts.md14.5 KiB
experts/bridge/ui-block-kit-adaptive-cards-ts.md21.5 KiB
experts/bridge/ui-legacy-attachments-cards-ts.md11.2 KiB
experts/bridge/ui-modals-dialogs-ts.md21.4 KiB
experts/bridge/workflow.composable-platform-ts.md13.7 KiB
experts/bridge/workflows-automation-ts.md14.3 KiB
experts/builder.md11.1 KiB
experts/convert/bulk-conversion-strategy-ts.md12.8 KiB
experts/convert/dependency-mapping-ts.md8.3 KiB
experts/convert/index.md3.5 KiB
experts/convert/java-to-ts-ts.md20.6 KiB
experts/convert/js-to-ts-ts.md6.2 KiB
experts/convert/json-serialization-ts.md11.0 KiB
experts/convert/kotlin-to-ts-ts.md11.2 KiB
experts/convert/ruby-to-ts-ts.md9.7 KiB
experts/convert/type-mapping-ts.md10.1 KiB
experts/deploy/aws-bot-deploy-ts.md14.6 KiB
experts/deploy/aws-cli-reference-ts.md52.4 KiB
experts/deploy/azure-bot-deploy-ts.md15.4 KiB
experts/deploy/azure-cli-reference-ts.md16.9 KiB
experts/deploy/index.md4.6 KiB
experts/fallback.md2.1 KiB
experts/index.md18.9 KiB
experts/models/anthropic-ts.md7.8 KiB
experts/models/bedrock-ts.md9.9 KiB
experts/models/foundry-cloud-ts.md8.1 KiB
experts/models/foundry-local-ts.md10.6 KiB
experts/models/index.md5.8 KiB
experts/models/openai-azure-openai-ts.md8.3 KiB
experts/models/oss-openai-compatible-ts.md10.7 KiB
experts/models/transformers-js-ts.md11.6 KiB
experts/prompt-engineer.md14.7 KiB
experts/researcher.md2.8 KiB
experts/security/index.md1.2 KiB
experts/security/input-validation-ts.md13.8 KiB
experts/security/secrets-ts.md13.9 KiB
experts/slack/bolt-assistant-ts.md9.9 KiB
experts/slack/bolt-events-ts.md9.3 KiB
experts/slack/bolt-java.md8.9 KiB
experts/slack/bolt-oauth-distribution-ts.md10.9 KiB
experts/slack/bolt-python.md9.5 KiB
experts/slack/cli.app-management.md7.9 KiB
experts/slack/cli.datastore-env.md8.6 KiB
experts/slack/cli.getting-started.md8.0 KiB
experts/slack/cli.local-dev-deploy.md8.4 KiB
experts/slack/cli.manifest-triggers.md9.9 KiB
experts/slack/index.md8.9 KiB
experts/slack/runtime.ack-rules-ts.md9.8 KiB
experts/slack/runtime.bolt-foundations-ts.md10.7 KiB
experts/slack/runtime.shortcuts-ts.md11.0 KiB
experts/slack/runtime.slash-commands-ts.md11.6 KiB
experts/slack/runtime.socket-mode-ts.md9.5 KiB
experts/slack/ui.block-kit-ts.md11.5 KiB
experts/slack/ui.modals-lifecycle-ts.md13.3 KiB
experts/slack/web-api-proactive-ts.md11.0 KiB
experts/slack/workflow.slack-automations-ts.md10.2 KiB
experts/teams/a2a.client-basics-ts.md9.7 KiB
experts/teams/a2a.orchestrator-patterns-ts.md12.7 KiB
experts/teams/a2a.server-basics-ts.md11.4 KiB
experts/teams/ai.chatprompt-basics-ts.md8.1 KiB
experts/teams/ai.citations-feedback-ts.md6.9 KiB
experts/teams/ai.conversational-query-ts.md14.2 KiB
experts/teams/ai.function-calling-design-ts.md9.0 KiB
experts/teams/ai.function-calling-implementation-ts.md7.7 KiB
experts/teams/ai.memory-localmemory-ts.md7.3 KiB
experts/teams/ai.model-setup-ts.md6.6 KiB
experts/teams/ai.rag-retrieval-ts.md7.4 KiB
experts/teams/ai.rag-vectorstores-ts.md8.5 KiB
experts/teams/ai.streaming-ts.md6.9 KiB
experts/teams/auth.oauth-sso-ts.md10.1 KiB
experts/teams/compat.botbuilder-interop-ts.md14.6 KiB
experts/teams/dev.debug-test-ts.md14.1 KiB
experts/teams/graph.usergraph-appgraph-ts.md9.0 KiB
experts/teams/index.md9.8 KiB
experts/teams/mcp.client-basics-ts.md9.5 KiB
experts/teams/mcp.expose-chatprompt-tools-ts.md9.5 KiB
experts/teams/mcp.security-ts.md10.4 KiB
experts/teams/mcp.server-basics-ts.md9.9 KiB
experts/teams/project.scaffold-files-ts.md12.2 KiB
experts/teams/runtime.app-init-ts.md10.9 KiB
experts/teams/runtime.manifest-ts.md12.3 KiB
experts/teams/runtime.proactive-messaging-ts.md12.6 KiB
experts/teams/runtime.routing-handlers-ts.md12.3 KiB
experts/teams/state.storage-patterns-ts.md9.4 KiB
experts/teams/teams-dotnet.md8.2 KiB
experts/teams/teams-python.md8.3 KiB
experts/teams/ui.adaptive-cards-ts.md12.9 KiB
experts/teams/ui.dialogs-task-modules-ts.md15.1 KiB
experts/teams/ui.message-extensions-ts.md15.6 KiB
experts/teams/workflow.approvals-inline-ts.md12.4 KiB
experts/teams/workflow.message-native-records-ts.md10.6 KiB
experts/teams/workflow.sharepoint-lists-ts.md9.0 KiB
experts/teams/workflow.state-driven-events-ts.md12.2 KiB
experts/teams/workflow.triggers-compose-ts.md10.3 KiB
experts/update-experts.md10.0 KiB
provision-deploy/provision-deploy.md4.3 KiB
test-playground/playground-cli.md4.6 KiB
test-playground/playground.md4.5 KiB
test-playground/test-playground.md1.1 KiB
test-teams/test-teams.md4.9 KiB
toolkit/README.md4.2 KiB
toolkit/commands.md2.6 KiB
toolkit/environments.md12.6 KiB
toolkit/lifecycle-cli.md17.6 KiB
toolkit/manifest-and-yaml.md6.0 KiB
toolkit/playground.md7.6 KiB
toolkit/publish.md9.3 KiB
toolkit/templates.md6.7 KiB
troubleshoot/troubleshoot.md20.2 KiB

experts/index.md

experts-router

purpose

You are the root task router. Before doing any work, interview the developer to understand scope and preferences. Then classify intent and load exactly one domain router. Do NOT load micro-expert files directly from here.

pre-task interview (mandatory)

Every task starts here. Before routing, loading experts, or writing any code, interview the developer. The depth scales with the task — a small bug fix may need one confirmation; a multi-file migration needs detailed scoping.

How it works

Assess complexity. Read the developer's message. Determine if the task is small (single file, clear intent), medium (multi-file, some ambiguity), or large (architectural, multi-step, cross-cutting).
Ask the right questions. Use AskUserQuestion to walk through the applicable question blocks below. Skip any question the developer's message already answers unambiguously.
Always offer the escape hatch. Every question set MUST include a "You decide everything" (first question) or "You decide everything else" (subsequent questions) option. Selecting it means: use your best judgment for all remaining decisions, apply expert defaults, and proceed without further questions.
Record and carry forward. Store all answers. They shape routing, expert selection, and implementation decisions for the entire task.

Question blocks

Pick from these based on complexity. You don't need all of them — use judgment.

Q1 — Scope & intent (always ask)

header: "Scope"
question: "Before I start, let me confirm what you need. Which best describes this task?"
options:
  - label: "You decide everything"
    description: "I trust your judgment — assess the task, make all decisions, and just do it."
  - label: "Quick fix / small change"
    description: "Single file or minor tweak. Just need it done."
  - label: "Feature / migration step"
    description: "Multi-file change with some decisions to make."
  - label: "Architectural / large task"
    description: "Significant scope — I want to review the approach before you start."
multiSelect: false

If the developer picks "You decide everything" → skip all remaining questions, proceed with expert defaults.

Q2 — Approach preferences (medium+ tasks)

header: "Approach"
question: "Any preferences on how I tackle this?"
options:
  - label: "You decide everything else"
    description: "No preferences — use best practices and expert defaults."
  - label: "Minimal changes"
    description: "Touch as few files as possible. Keep the diff small."
  - label: "Do it right"
    description: "Refactor if needed. Prioritize correctness and clean code."
  - label: "Let me review first"
    description: "Show me a plan before writing any code."
multiSelect: false

If "Let me review first" → enter plan mode (EnterPlanMode) before implementation.

Q3 — Specifics (large tasks or when ambiguity remains)

header: "Details"
question: "A few specifics to nail down:"
options:
  - label: "You decide everything else"
    description: "Use your best judgment for all remaining decisions."
  - label: "I'll answer each"
    description: "Walk me through the decisions one at a time."
multiSelect: false

If "I'll answer each" → proceed to any expert-level ## interview sections after routing. If "You decide everything else" → fill expert interviews with defaults and skip them.

Interaction with expert-level interviews

The pre-task interview gates the entire workflow. Expert-level ## interview sections (described in the auto-interview protocol below) handle domain-specific decisions. If the developer chose "You decide everything" or "You decide everything else" at the pre-task level, those expert interviews are auto-filled with defaults and skipped.

routing rules

Scan the user's message for signal words. Pick the first matching domain; if signals overlap, prefer the domain whose signals appear more often.

Teams — build or modify a Teams bot / agent

Signals: Teams SDK, @microsoft/teams-ai, Adaptive Cards, ChatPrompt, Graph API, MCP, A2A, app.ts, manifest, proactive message, message extension, dialog, task module, Agents Toolkit, Bot Framework, SSO, OAuth, streaming, citations, RAG, function calling, memory, state, storage, microsoft_teams, microsoft_teams.apps, microsoft_teams.ai, ActivityContext, @app.on_message, OpenAICompletionsAIModel, Microsoft.Teams.Apps, Microsoft.Teams.AI, AddTeams(), UseTeams(), IContext<TActivity>, OnMessage, OpenAIChatPrompt, Teams Python, Teams .NET, Teams C#, m365agents.yml, atk provision, atk deploy, atk publish, atk preview, atk validate, atk package, atk new, Agents Playground, .m365agentsplayground.yml, lifecycle hooks, env/.env → Read experts/teams/index.md

Slack — build or modify a Slack app

Signals: Slack Bolt, @slack/bolt, Block Kit, ack(), Slack events, app.message, app.command, app.event, app.action, app.shortcut, app.view, slash command, Slack OAuth, InstallProvider, InstallationStore, multi-workspace, app.assistant, Assistant container, threadStarted, userMessage, setSuggestedPrompts, setStatus, setTitle, getThreadContext, Socket Mode, socketMode, appToken, @slack/socket-mode, SocketModeReceiver, chat.postMessage, chat.update, chat.postEphemeral, proactive message, WebClient, app.client, filesUploadV2, global shortcut, message shortcut, message_action, modal, views.open, views.update, views.push, view_submission, view_closed, private_metadata, slack_bolt, AsyncApp, SocketModeHandler, chat_postMessage, views_open, slack-bolt-java, com.slack.api.bolt, AppConfig.builder(), MethodsClient, Slack Python, Slack Java, Slack CLI, slack create, slack run, slack deploy, slack activity, slack trigger, slack datastore, slack env, slack manifest, slack auth login, slack doctor, slack app install, slack collaborator, slack function distribute, slack project, .slack/, project.json, manifest.ts, slack.json, trigger definition, Slack hosted platform → Read experts/slack/index.md

Bridge — cross-platform bridging between Slack and Teams, or AWS and Azure

Signals: bridge, cross-platform, migrate, migration, convert, port, Slack to Teams, Teams to Slack, Slack→Teams, Teams→Slack, Slack↔Teams, Block Kit to Adaptive Cards, Adaptive Cards to Block Kit, AWS to Azure, Azure to AWS, Lambda to Functions, S3 to Blob, CloudWatch to App Insights, middleware chain, ack(), Socket Mode, RTM, transport, WebSocket to HTTPS, InstallationService, OAuthStateService, views.open, viewSubmission, viewClosed, App Home, views.publish, attachmentAction, legacy attachments, callback_id, modals to dialogs, task module, replace_original, delete_original, chat.update, chat.postEphemeral, response_url, ephemeral, files.upload, file_shared, link_shared, chat.unfurl, unfurl, shortcut, global shortcut, message shortcut, scheduleMessage, reminders.add, conversations.create, conversations.archive, conversations.invite, channel ops, Workflow Builder, workflow_step_execute, Power Automate, App Directory, InstallationStore, sideloading, rate limit, retry, 429, throttle, circuit breaker, Dapr, FileConsentCard, add Teams support, add Slack support, help me migrate, cross-platform advisor, bridging decisions, assess bridging, bridging scope, dual bot, single server, shared Express, REST API, Java, C#, Go, no SDK, raw HTTP, Bot Framework REST, manual JWT, signature verification, Python dual-platform, Python cross-platform, slack_bolt + microsoft_teams, unified Python server → Read experts/bridge/index.md

Models — configure and call AI models from any provider

Signals: OpenAI, Azure OpenAI, GPT-4o, GPT-4, Anthropic, Claude, @anthropic-ai/sdk, Bedrock, @aws-sdk/client-bedrock-runtime, Converse API, Foundry Local, foundry model, foundry service, Foundry cloud, MaaS, model-as-a-service, GitHub Models, Ollama, vLLM, LM Studio, llama.cpp, TGI, LocalAI, open-source model, local LLM, OpenAI-compatible, /v1/chat/completions, openai npm, model provider, AI model, LLM, chat completions, tool use, function calling, Bedrock agents, Bedrock guardrails, Phi-4, Qwen, Llama, Mistral, DeepSeek, embeddings, model selection, Transformers.js, @huggingface/transformers, in-process inference, browser inference, WASM, WebGPU, local embeddings, local classification, pipeline API, HuggingFace → Read experts/models/index.md

Deploy — deploy a bot to Azure or AWS

Signals: deploy, deployment, provision, hosting, publish, go live, production, az login, aws configure, App Service, Azure Functions, Container Apps, Lambda, EC2, ECS, Fargate, Elastic Beanstalk, Azure Bot, atk provision, atk deploy, Agents Toolkit deploy, zip deploy, SAM deploy, CDK deploy, API Gateway, deploy to Azure, deploy to AWS, push to cloud, CloudFormation → Read experts/deploy/index.md

Convert — rewrite source code from another language to TypeScript

Signals: JS to TS, Ruby to TS, Java to TS, Kotlin to TS, convert language, transpile, rewrite in TypeScript, port code, language conversion, gems to npm, Maven to npm, Gradle to npm, type annotations, require to import, add types, snake_case to camelCase, Lombok, @Data, @Builder, Gson, Jackson, @SerializedName, CompletableFuture, bulk conversion, large-scale, trailing lambda, it parameter, trimIndent, when expression, data class, companion object, sealed class, extension function → Read experts/convert/index.md

Security — harden inputs, secrets, or credentials

Signals: input validation, sanitize, secrets, credentials, key vault, PII, injection, XSS → Read experts/security/index.md

auto-interview protocol (expert-level)

After loading any expert, check for a ## interview section. If one exists and the developer did NOT choose "You decide everything" / "You decide everything else" in the pre-task interview, execute the expert interview before writing any code or giving implementation advice.

If the developer DID choose an escape hatch in the pre-task interview, auto-fill all expert interview answers with the expert's documented defaults and skip the questions.

How it works

Check pre-task answers. If the developer already opted out of detailed questions, skip to implementation using defaults.
Detect. After reading an expert file, scan for ## interview. If missing, proceed directly to implementation.
Execute. Walk through each question block (### Q1, ### Q2, ...) in order. Use AskUserQuestion for each one, following the header, question, options, and multiSelect fields exactly as specified.
Record. Store the developer's answers. Pass them into the expert's rules and patterns as context — the answers shape which code paths, patterns, and configurations apply.
Escape hatch. Every expert interview MUST include a "You decide everything else" option. Selecting it fills all remaining answers with the expert's documented defaults and skips remaining questions.
Context-skip. If the developer's original message already answers an interview question unambiguously (e.g., "use Azure OpenAI" when the question is "OpenAI or Azure OpenAI?"), skip that question — don't re-ask what's already known.

Experts with embedded interviews

Some experts (like cross-platform-advisor-ts.md) have the interview woven into their phased workflow rather than in a separate ## interview section. These already satisfy the protocol — their Phase 2 / decision walkthrough IS the interview.

ambiguity tiebreaker

If the request mixes bridge signals with a target-platform signal (e.g., "convert Block Kit to Adaptive Cards"), route to bridge first — bridge experts already reference the target platform's patterns.

If the request mixes convert signals with bridge signals (e.g., "rewrite this Ruby Slack bot in TypeScript for Teams"), route to convert first for language translation, then bridge for platform mapping. The convert domain's combining rule handles this layering.

fallback

If no domain matches, ask one clarifying question:

"Are you working on a Teams bot, a Slack app, a migration between them, or something else?"

If the routed experts don't fully cover the request (gaps remain after the initial pass), read experts/fallback.md for a two-phase recovery: re-scan all domain routers for missed experts, then web-search for any remaining gaps.

expert evolution

The expert system is self-evolving. As conversations reveal knowledge gaps, outdated patterns, or new topic areas, update the system in-place. Follow the rules below.

Naming conventions

Convention	Meaning
`topic-ts.md`	Normal expert — can be created, updated, or replaced
`topic-ts.locked.md`	Locked — read-only, do NOT edit. Rename to remove `.locked` only with explicit user approval
`_filename.md`	System/template file — not a routable expert
`index.md`	Domain router — update when adding/removing experts in that domain

To lock an expert: rename topic-ts.md → topic-ts.locked.md. The routing system treats .locked.md identically for reads — only writes are blocked.

When to UPDATE an existing expert

Update an expert when any of these are true during a conversation:

Corrected mistake — you discover a rule, code pattern, or pitfall is wrong and fix it in practice. Backport the correction to the expert.
New pattern emerged — you write a working pattern that isn't covered by any existing expert. Add it to the most relevant expert's ## patterns section.
SDK/API changed — the user provides or you discover updated API signatures, new options, or deprecated features. Update the affected rules and patterns.
Missing pitfall — you hit a gotcha during implementation that the expert didn't warn about. Add it to ## pitfalls.
Cross-reference gap — an expert should reference another but doesn't. Add a Pair with entry to ## instructions and update the domain index.md cluster.

Do NOT update locked experts (*.locked.md). If a locked expert needs changes, flag it to the user.

When to CREATE a new expert

Create a new expert when all of these are true:

No existing expert covers the topic — the knowledge doesn't fit as an addition to any current file.
The topic is reusable — it will apply to future tasks, not just the current one-off request.
Sufficient depth — the topic warrants 8+ rules and 2+ code patterns. If it's only 2-3 rules, add them to an existing expert instead.

Creation steps:

Copy the template from experts/_expert-ts.md.
Name it {topic}-ts.md in the appropriate domain folder.
Fill in all sections: purpose, rules, patterns, pitfalls, references, instructions (with Pair with cross-refs), research.
Complete the post-creation checklist from _expert-ts.md:
- Add to domain index.md task cluster Read: list (with Depends on: / Cross-domain deps: if applicable).
- Add to domain index.md file inventory.
- Add signal words to root index.md if the new expert introduces terms not already covered.

When to CREATE a new domain folder

Create a new domain (folder + index.md router) when all of these are true:

3+ experts would belong to the new domain — a domain with 1-2 experts should stay as a cluster within an existing domain.
Distinct signal words — the domain's topics wouldn't naturally route through any existing domain's signals.
Separable routing — moving these experts out of an existing domain simplifies that domain's router, not complicates it.

Creation steps:

Create the folder: experts/{domain-name}/.
Create experts/{domain-name}/index.md following the router pattern from any existing domain index (purpose, task clusters with When:/Read:, combining rule, file inventory).
Move or create the expert files in the new folder.
Add a new routing entry to this root index.md under ## routing rules with Signals: and a → Read directive.
Remove any moved experts from their old domain's index.md.

Evolution audit trail

When you modify the expert system, add a one-line comment at the bottom of the affected domain's index.md:

<!-- Updated YYYY-MM-DD: {what changed and why} -->

This lets future sessions see what evolved and when, without cluttering the routing logic.

utilities

These files support the expert system itself — not a specific domain.

Fallback recovery

→ Read experts/fallback.md When the initial routing pass leaves knowledge gaps.

New expert template

→ Read experts/_expert-ts.md When creating a new micro-expert file. Provides the canonical stub structure and a post-creation checklist.

Research workflow

→ Read experts/researcher.md When fleshing out a stub expert with real content. Provides the step-by-step Deep Research workflow.

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Needs an API key, Needs Python, Needs Node.js

Security

What our scanner noticed· Medium risk

Possible data-exfiltration endpoints
Instructions that use the network
Downloads piped into a shell

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Published by

Microsoft

Publisher

Microsoft (Official)

Source

microsoft/skills · .github/plugins/microsoft-365-agents-toolkit/skills/teams-app-developer · commit 0696600

Licence

MIT License (redistribution allowed)

LICENSE (added by AutoGineer from LICENSE)
ATTRIBUTION.md (written by AutoGineer)

Synced

2026-11-09

AutoGineer is not affiliated with or endorsed by Microsoft or Microsoft.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download teams-app-developer.zip

From the author

AutoGineer summary

What it does

When to use it

Advantages

Usage examples

Install

Documentation by Microsoft

Microsoft 365 Agents Toolkit Skill

AI Behavior Guidelines

ATK CLI Setup

CLI Global Options

Sub-Skills

Shared References

Workflow Chains

ATK Project Context Resolution

Step 1: Detect ATK Project

Step 2: Resolve Common Configuration

Step 3: Collect Missing Values

Files

experts/index.md

experts-router

purpose

pre-task interview (mandatory)

How it works

Question blocks

Q1 — Scope & intent (always ask)

Q2 — Approach preferences (medium+ tasks)

Q3 — Specifics (large tasks or when ambiguity remains)

Interaction with expert-level interviews

routing rules

Teams — build or modify a Teams bot / agent

Slack — build or modify a Slack app

Bridge — cross-platform bridging between Slack and Teams, or AWS and Azure

Models — configure and call AI models from any provider

Deploy — deploy a bot to Azure or AWS

Convert — rewrite source code from another language to TypeScript

Security — harden inputs, secrets, or credentials

auto-interview protocol (expert-level)

How it works

Experts with embedded interviews

ambiguity tiebreaker

fallback

expert evolution

Naming conventions

When to UPDATE an existing expert

When to CREATE a new expert

When to CREATE a new domain folder

Evolution audit trail

utilities

Fallback recovery

New expert template

Research workflow

Requirements and compatibility

Security

Attribution and licence

Related skills

Implement Spec

MCP Builder

Schema

Postgresql Table Design

SQL Optimization Patterns

API Scaffolding Fastapi Pro

Download ZIP