analyze-objects — Claude Code skill for object analysis
analyze-objects is a Claude Code skill for studying processes with interacting object types and delays across objects. It activates for questions about leading-object executions or process paths.
MIT ★ 1,167 updated 2026-10-29
/plugin marketplace add microsoft/power-platform-skills
/plugin install process-intelligence@power-platform-skillsWhat analyze-objects does
analyze-objects retrieves execution statistics for a selected leading object type, then gathers the requested evidence: executions, objects, or edges. It first fetches available object types and asks for clarification when the goal does not identify one clearly. With a cached type and scope, it can go straight to a specific drilldown.
The skill keeps object filters separate from case filters and makes the leading type, filter propagation, and date mode explicit. Reports include native units and page coverage. The requested OCPM tools must be available; otherwise, the workflow stops rather than switching to case analytics. A delay claim also requires a verified business definition and threshold.
What it helps with
- count process executions for a leading object type
- retrieve objects or edges matching selected filters
- inspect a path from its first event to its last
- test an object delay against a defined threshold
Example prompts for the agent
- Analyze process executions for the leading object type Order.
- Find object relationships within the selected date range.
- Check the delay from the first Created event to the last Paid event.
How to install analyze-objects
1Plugin marketplace
/plugin marketplace add microsoft/power-platform-skills/plugin install process-intelligence@power-platform-skillsThis installs the whole process-intelligence plugin with all its skills
2Skills CLI (Vercel)
npx skills add microsoft/power-platform-skills --skill analyze-objects -a claude-code3Manually: put the files in place
~/.claude/skills/analyze-objects/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/analyze-objects/
1Skills CLI (Vercel)
npx skills add microsoft/power-platform-skills --skill analyze-objects -a codex2Manually: put the files in place
.agents/skills/analyze-objects/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add microsoft/power-platform-skills --skill analyze-objects -a cursor2Manually: put the files in place
.cursor/skills/analyze-objects/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install microsoft/power-platform-skills plugins/process-intelligence/skills/analyze-objects2Manually: put the files in place
.github/skills/analyze-objects/SKILL.md1Skills CLI (Vercel)
npx skills add microsoft/power-platform-skills --skill analyze-objects -a gemini-cli2Platform CLI
gemini skills install https://github.com/microsoft/power-platform-skills --path plugins/process-intelligence/skills/analyze-objectsManually: put the files in place
.devin/skills/analyze-objects/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about analyze-objects
How do I install analyze-objects in Claude Code?
Add the marketplace with /plugin marketplace add microsoft/power-platform-skills, then install the plugin with /plugin install process-intelligence@power-platform-skills.
What does analyze-objects do?
It examines processes involving multiple object types. The skill can count executions and retrieve requested executions, objects, or relationships.
What does analyze-objects need to run?
It requires the OCPM tools needed for the requested analysis. If a required tool is unavailable, the workflow stops instead of falling back to case analytics.
Source text
SKILL.md · MIT
---
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:
```text
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:
```text
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.
© Microsoft · MIT · microsoft/power-platform-skills