Skip to content
Development
Skill

/analyze-objects

Use when analyzing interacting object types, object-centric processes, leading-object executions or cross-object delay.

BOOST
From plugin
power-platform-skills
987107 skills24 agents5 MCP
Install
$ npx -y skills add microsoft/power-platform-skills --skill analyze-objects --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/analyze-objects

Context preview

The summary Claude sees to decide when to auto-load this skill.

Use when analyzing interacting object types, object-centric processes, leading-object executions or cross-object delay.

SKILL.md

analyze-objects.SKILL.md
name: analyze-objects
description: Use when analyzing interacting object types, object-centric processes, leading-object executions or cross-object delay.
allowed-tools: Read, AskUserQuestion

Object-centric analysis, not relabeled cases

Use the [shared contract](../../references/analysis-contract.md). Require the **needed** advertised OCPM tools; absence stops this workflow, never falls back to case analytics. Keep separate object context and `filters:Fobj`; do not copy case `filterOptions`.

Ordered trace

1. Only if unknown: `get_ocpm_processes {}` -> select returned P -> `get_process_details_v2 {processId:P}` for missing metadata. Case `get_processes` does not discover object-centric processes. 2. Unless cached: `get_ocpm_object_types {S,itemsPerPage:20,itemsToSkip:0}`. Choose **returned** leading type O from the user's objective; ask if ambiguous. 3. `get_ocpm_process_execution_statistics {S,leadingObjectTypeName:O,filters:Fobj}`. `ProcessExecutionCount` counts executions; `ProcessExecutionDuration` is their mean. 4. Choose the requested evidence branch: `get_ocpm_process_executions {S,leadingObjectTypeName:O,filters:Fobj,itemsPerPage:5,itemsToSkip:0}`, or `get_ocpm_objects {S,objectTypeName:O,filters:Fobj,itemsPerPage:5,itemsToSkip:0}`, or `get_ocpm_edges {S,filters:Fobj,itemsPerPage:10,itemsToSkip:0}`.

An exact object/execution request with cached type/scope can go directly to its drilldown; do not force a new summary. Continue needed pages or discriminating follow-ups without fixed call/row/drilldown quotas. Reuse complete results; stop when answered or no new evidence remains. No OCPM sort, generic metrics/grouping or formula-column inputs exist. Search is supported only on object types, objects and edges; aggregate statistics has no paging.

Filter construction

Use exact returned type names, attributes and object identifiers (objects expose `Name`, not invented GUIDs). Declare leading type, propagation and date mode. For a selected-type-only timeframe:

objectTimeframeFilters:[{objectTypeName:O,startDate:START,endDate:END,
  filteringType:"Contained",isInclusive:true,propagation:"SelectedObjectTypeOnly"}]

SelectedObjectTypeOnly forbids propagationDepth. RelatedObjects defaults to depth 2; depth N means N-1 relation hops. Choose explicitly when ambiguity changes scope. OCPM timeframe requires start before end and describes an exclusive end; unlike case filters, its discovered modes can include Started, Completed, Intersecting and PassThrough.

For a first-to-last path threshold with known activity values FROM/TO:

processExecutionFilters:[{leadingObjectTypeName:O,
  from:{selectionType:"First",value:FROM},to:{selectionType:"Last",value:TO},
  timeBetweenEventsRelation:"GreaterThan",timeBetweenEventsSeconds:SECONDS,isInclusive:true}]

Time relation and finite nonnegative seconds must be paired. Secondary object-metric Duration conditions instead use **ticks**; case duration filters use TimeSpan strings. Do not interchange them.

Report at most five headline rows with native units, leading type, filters and page coverage. Object-type counts are not additive unique executions. Execution Duration is not automatically the filtered path interval or invoice-to-order delay. A delay claim also needs a verified business definition/threshold. Mark unsupported interpretations **inconclusive**, not inferred from cases. Use the [investigation method](../../references/investigation-method.md) for evidence labels and compact references, counts, units, coverage and alternatives. Freeze baseline object filters separately from test filters; keep object context, leading type and propagation explicit. Stop on missing interval evidence or capability, never substitute a case-based model.

Read more
Ships withpower-platform-skills

Official agent skills/plugins for Power Platform development by Microsoft.

Get the whole plugin

Other skills on power-platform-skills.