int-reference
Skill486 starsChanged 8 months agoMITView on GitHub
Reference tables for Copilot Studio YAML authoring: triggers, actions, variables, entities, Power Fx functions, templates. Preloaded by author and advisor agents.
In this repository30 other files it gives its agents.
- add-actionSkill
- add-adaptive-cardSkill
- add-generative-answersSkill
- add-global-variableSkill
- add-knowledgeSkill
- add-nodeSkill
- add-other-agentsSkill
- analyze-evalsSkill
- chat-directlineSkill
- chat-sdkSkill
- chat-with-agentSkill
- clone-agentSkill
- create-eval-setSkill
- create-evalSkill
- detect-modeSkill
- directline-chatSkill
- edit-actionSkill
- edit-agentSkill
- edit-triggersSkill
- int-patternsSkill
- int-project-contextSkill
- list-kindsSkill
- list-topicsSkill
- lookup-schemaSkill
- manage-agentSkill
- new-topicSkill
- run-evalSkill
- run-tests-kitSkill
- test-authSkill
- validateSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 128
Never mix these two formats.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Copilot Studio YAML Reference
Core File Types
| File | Purpose |
agent.mcs.yml | Main agent metadata (kind: GptComponentMetadata) |
settings.mcs.yml | Agent settings and configuration |
connectionreferences.mcs.yml | Connector references |
topics/*.mcs.yml | Conversation topics (kind: AdaptiveDialog) |
actions/*.mcs.yml | Connector-based actions (kind: TaskDialog) |
knowledge/*.mcs.yml | Knowledge sources (kind: KnowledgeSourceConfiguration) |
variables/*.mcs.yml | Global variables (kind: GlobalVariableComponent) |
agents/*.mcs.yml | Child agents (kind: AgentDialog) |
Trigger Types
Topics with OnRecognizedIntent have two routing mechanisms — which one matters depends on the orchestration mode:
modelDescription— used by generative orchestration (GenerativeActionsEnabled: true). The AI orchestrator reads this to decide routing. Primary mechanism for generative agents.- Trigger phrases (
triggerQueries) — used by classic orchestration. Pattern-matched against the user's utterance. Secondary hints when generative orchestration is enabled.
System triggers (OnConversationStart, OnUnknownIntent, OnError, etc.) fire automatically and don't use either mechanism.
| Kind | Purpose |
OnRecognizedIntent | Trigger phrases matched |
OnConversationStart | Conversation begins |
OnUnknownIntent | No topic matched (fallback) |
OnEscalate | User requests human agent |
OnError | Error handling |
OnSystemRedirect | Triggered by redirect only |
OnSelectIntent | Multiple topics matched (disambiguation) |
OnSignIn | Authentication required |
OnToolSelected | Child agent invocation |
OnKnowledgeRequested | Custom knowledge source search triggered (YAML-only, no UI) |
OnGeneratedResponse | Intercept AI-generated response before sending |
OnOutgoingMessage | Non-functional (2026-03-15) — exists in schema but does not fire at runtime. Do not use. |
YAML-Only Features
These features work at runtime but are not visible in the Copilot Studio UI. Warn users that UI edits may silently remove them.
| Feature | Notes |
triggerCondition on knowledge sources | The UI only exposes this as an on/off toggle (=false to exclude from UniversalSearchTool). Arbitrary Power Fx expressions (e.g., =Global.UserDepartment = "HR") work at runtime but can only be set via YAML. Use with caution. (2026-03-16) |
Action Types
| Kind | Purpose |
SendActivity | Send a message |
Question | Ask user for input |
SetVariable | Set/compute a variable (Power Fx expression, prefix =) |
SetTextVariable | Set a text variable using template interpolation ({}). Useful for converting non-text types (e.g., Number) to text: "You have {Topic.Count} items" |
ConditionGroup | Branching logic |
BeginDialog | Call another topic |
ReplaceDialog | Replace current topic |
EndDialog | End current topic |
CancelAllDialogs | Cancel all topics |
ClearAllVariables | Clear variables |
SearchAndSummarizeContent | Generative answers (grounded in knowledge) |
AnswerQuestionWithAI | AI answer (conversation history + general knowledge only) |
EditTable | Modify a collection |
CSATQuestion | Customer satisfaction |
LogCustomTelemetryEvent | Logging |
OAuthInput | Sign-in prompt |
SearchKnowledgeSources | Search knowledge sources (returns raw results, no AI summary) |
CreateSearchQuery | AI-generated search query from user input |
Connector Actions (TaskDialog)
Connector actions (kind: TaskDialog) invoke external connector operations. They are stored in actions/ and require a connection reference in connectionreferences.mcs.yml.
Use /add-action to create new actions from available connectors. The schema describes the structural properties of TaskDialog and InvokeConnectorTaskAction, but the specific inputs and outputs for each connector operation are connector-specific — use the connector lookup script (connector-lookup.bundle.js) to get the full operation details.
Action Structure
| Field | Purpose |
kind: TaskDialog | Identifies this as a connector action |
inputs | Inputs: AutomaticTaskInput (AI-provided) or ManualTaskInput (fixed value) |
modelDisplayName | Display name for AI orchestrator routing |
modelDescription | Description for AI orchestrator routing |
outputs | Output property names returned by the connector |
action.kind | Always InvokeConnectorTaskAction for connector actions |
action.connectionReference | Logical name of the connection (registered in connectionreferences.mcs.yml) |
action.connectionProperties.mode | Maker (maker's credentials) or Invoker (end user's credentials) |
action.operationId | The connector's specific operation identifier |
outputMode | Usually All — exports all operation outputs |
Input Types
| Input Kind | Use When | Notes |
AutomaticTaskInput | The AI orchestrator should provide the value based on context | Includes description for the AI to understand what to provide |
ManualTaskInput | A fixed/hardcoded value (e.g., timezone, folder path) | Can only hardcode strings. Non-string values (IDs, enums) should be reviewed by the user after pushing |
$-Prefixed Property Names (SharePoint, OData)
Some connectors (notably SharePoint) use OData parameters like $filter, $orderby, $top. These require special quoting in TaskDialog YAML — both single and double quotes:
# TaskDialog (actions/*.mcs.yml) — CORRECT
- kind: ManualTaskInput
propertyName: "'$filter'"
value: "Status eq 'Active'""'$filter'" means: the outer "" are YAML string delimiters; the inner '' are part of the literal value sent to the runtime. Using $filter, "$filter", or '$filter' alone will fail.
InvokeConnectorAction (inline in topics) uses a different format — the parameters/ prefix with no inner single quotes:
# InvokeConnectorAction (inside topics) — CORRECT
- kind: InvokeConnectorAction
operationId: GetItems
input:
parameters/$filter: "Status eq 'Active'"Never mix these two formats.
System Variables
| Variable | Description |
System.Bot.Name | Agent's name |
System.Activity.Text | User's current message |
System.Conversation.Id | Conversation identifier |
System.Conversation.InTestMode | True if in test chat |
System.FallbackCount | Number of consecutive fallbacks |
System.Error.Message | Error message |
System.Error.Code | Error code |
System.SignInReason | Why sign-in was triggered |
System.Recognizer.IntentOptions | Matched intents for disambiguation |
System.Recognizer.SelectedIntent | User's selected intent |
System.SearchQuery | AI-rewritten search query (available in OnKnowledgeRequested) |
System.KeywordSearchQuery | Keyword version of search query (available in OnKnowledgeRequested) |
System.SearchResults | Table to populate with custom search results — schema: Content, ContentLocation, Title (available in OnKnowledgeRequested) |
System.ContinueResponse | Set to false in OnGeneratedResponse to suppress auto-send |
System.Response.FormattedText | The AI-generated response text (available in OnGeneratedResponse) |
Variable Scopes
| Prefix | Scope | Lifetime |
Topic.<name> | Topic variable | Current topic only |
Global.<name> | Global variable | Entire conversation (defined in variables/ folder) |
System.<name> | System variable | Built-in, read-only |
Global variables are defined as YAML files in variables/<Name>.mcs.yml (kind: GlobalVariableComponent). aIVisibility accepts UseInAIContext (orchestrator can read and reason about the value) or Hidden (orchestrator unaware — use for flags and internal bookkeeping).
Prebuilt Entities
| Entity | Use Case |
BooleanPrebuiltEntity | Yes/No questions |
NumberPrebuiltEntity | Numeric inputs |
StringPrebuiltEntity | Free text |
DateTimePrebuiltEntity | Date/time |
EMailPrebuiltEntity | Email addresses |
Power Fx Expression Reference
Only use functions from the supported list below. Copilot Studio supports a subset of Power Fx — using unsupported functions will cause errors.
# Arithmetic
value: =Text(Topic.number1 + Topic.number2)
# Date formatting
value: =Text(Now(), DateTimeFormat.UTC)
# Conditions
condition: =System.FallbackCount < 3
condition: =Topic.EndConversation = true
condition: =!IsBlank(Topic.Answer)
condition: =System.Conversation.InTestMode = true
condition: =System.SignInReason = SignInReason.SignInRequired
condition: =System.Recognizer.SelectedIntent.TopicId = "NoTopic"
# String interpolation in activity (uses {} without =)
activity: "Error: {System.Error.Message}"
activity: "Error code: {System.Error.Code}, Time (UTC): {Topic.CurrentTime}"
# Record creation
value: "={ DisplayName: Topic.NoneOfTheseDisplayName, TopicId: \"NoTopic\", TriggerId: \"NoTrigger\", Score: 1.0 }"
# Variable initialization (first assignment uses init: prefix)
variable: init:Topic.UserEmail
variable: init:Topic.CurrentTime
# Subsequent assignments omit init:
variable: Topic.UserEmailSupported Power Fx Functions
These are all the Power Fx functions available in Copilot Studio. Do NOT use any function not on this list.
Math: Abs, Acos, Acot, Asin, Atan, Atan2, Cos, Cot, Degrees, Exp, Int, Ln, Log, Mod, Pi, Power, Radians, Rand, RandBetween, Round, RoundDown, RoundUp, Sin, Sqrt, Sum, Tan, Trunc
Text: Char, Concat, Concatenate, EncodeHTML, EncodeUrl, EndsWith, Find, Left, Len, Lower, Match, MatchAll, Mid, PlainText, Proper, Replace, Right, Search, Split, StartsWith, Substitute, Text, Trim, TrimEnds, UniChar, Upper, Value
Date/Time: Date, DateAdd, DateDiff, DateTime, DateTimeValue, DateValue, Day, EDate, EOMonth, Hour, IsToday, Minute, Month, Now, Second, Time, TimeValue, TimeZoneOffset, Today, Weekday, WeekNum, Year
Logical: And, Coalesce, If, IfError, IsBlank, IsBlankOrError, IsEmpty, IsError, IsMatch, IsNumeric, IsType, Not, Or, Switch
Table: AddColumns, Column, ColumnNames, Count, CountA, CountIf, CountRows, Distinct, DropColumns, Filter, First, FirstN, ForAll, Index, Last, LastN, LookUp, Patch, Refresh, RenameColumns, Sequence, ShowColumns, Shuffle, Sort, SortByColumns, Summarize, Table
Aggregate: Average, Max, Min, StdevP, VarP
Type conversion: AsType, Boolean, Dec2Hex, Decimal, Float, GUID, Hex2Dec, JSON, ParseJSON
Other: Blank, ColorFade, ColorValue, Error, Language, OptionSetInfo, RGBA, Trace, With
Available Templates
Templates are bundled with the plugin. Skills that use templates reference them via ${CLAUDE_SKILL_DIR}/../../templates/.
| Template | File | Pattern |
| Greeting | templates/topics/greeting.topic.mcs.yml | OnConversationStart welcome |
| Fallback | templates/topics/fallback.topic.mcs.yml | OnUnknownIntent with escalation |
| Arithmetic | templates/topics/arithmeticsum.topic.mcs.yml | Inputs/outputs with computation |
| Question + Branching | templates/topics/question-topic.topic.mcs.yml | Question with ConditionGroup |
| Knowledge Search | templates/topics/search-topic.topic.mcs.yml | SearchAndSummarizeContent fallback |
| Custom Knowledge Source | templates/topics/custom-knowledge-source.topic.mcs.yml | OnKnowledgeRequested with custom API (YAML-only) |
| Remove Citations | templates/topics/remove-citations.topic.mcs.yml | OnGeneratedResponse citation stripping |
| Authentication | templates/topics/auth-topic.topic.mcs.yml | OnSignIn with OAuthInput |
| Error Handler | templates/topics/error-handler.topic.mcs.yml | OnError with telemetry |
| Disambiguation | templates/topics/disambiguation.topic.mcs.yml | OnSelectIntent flow |
| Agent | templates/agents/agent.mcs.yml | GptComponentMetadata |
| Connector Action (generic) | templates/actions/connector-action.mcs.yml | TaskDialog with connector (structural reference) |
| Knowledge (Public Website) | templates/knowledge/public-website.knowledge.mcs.yml | PublicSiteSearchSource |
| Knowledge (SharePoint) | templates/knowledge/sharepoint.knowledge.mcs.yml | SharePointSearchSource |
| Global Variable | templates/variables/global-variable.variable.mcs.yml | GlobalVariableComponent |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

