By name: HigherEdUnitResearch description: Research and profile organizational units within higher education institutions worldwide. Given criteria (unit type, institution filters), discovers matching units, researches each across 7 dimensions, and produces comprehensive profiles plus comparative analysis. USE WHEN user says "research [unit type] at [universities]", "profile [center/office/institute]", "compare [units] across [institutions]", "find AI centers at R1 universities", "who runs [unit] at [university]", or any request to systematically research higher education organizational units. metadata: author: dale-pike version: "1.0" category: research
HigherEdUnitResearch
Research and profile organizational units within higher education institutions. Discovers units matching criteria, researches each across 7 structured dimensions, and produces individual profiles, leader profiles, and comparative analyses.
Capability Detection
At workflow start, detect the execution environment to determine available capabilities:
CHECK: Does ~/.claude/skills/ exist?
YES -> PAI environment detected
CHECK: Are researcher agents available? (perplexity-researcher, claude-researcher, gemini-researcher)
YES -> Tier 1: Full PAI (parallel researcher fleet)
NO -> Tier 2: Partial PAI (limited researchers)
NO -> Tier 3: Bare Claude Code (WebSearch + WebFetch only)
| Tier | Researchers | Browser | Document Harvest | Parallelism |
|---|---|---|---|---|
| Tier 1: Full PAI | All *-researcher agents | Browser skill | PaperProcessor if available | 7+ simultaneous agents |
| Tier 2: Partial PAI | Available researchers | WebFetch fallback | Skip | 3-5 simultaneous agents |
| Tier 3: Bare Claude Code | WebSearch + WebFetch | WebFetch only | Skip | Sequential research |
The skill works at all three tiers. Lower tiers are slower but produce the same output structure.
Workflow Routing
| User Intent | Workflow | Depth Default |
|---|---|---|
| Research a single named unit | Workflows/SingleUnit.md |
standard |
| Compare multiple units or "find units matching criteria" | Workflows/MultiUnit.md |
standard |
| Refresh or update an existing profile | Workflows/UnitUpdate.md |
(targets stale dimensions) |
| Quick lookup ("who runs X at Y") | Workflows/SingleUnit.md |
quick |
| Deep dive on a specific unit | Workflows/SingleUnit.md |
deep |
Depth can be overridden by user language:
- "quick" / "brief" / "overview" / "scan" -> quick
- "deep" / "thorough" / "comprehensive" / "deep dive" -> deep
- No qualifier -> standard
Research Dimensions
Every unit is profiled across 7 structured dimensions (see UnitProfileSchema.md):
- Charter/Mission - Founding context, current purpose, strategic alignment
- Leadership - Director, governance, leadership team
- Organizational Position - Reporting line, unit type, peer units
- Services & Programs - What the unit does, for whom, how
- Key Artifacts - Website, reports, frameworks, publications
- Staffing & Scale - Staff estimates, budget indicators, growth
- External Recognition - Awards, partnerships, conference presence
Resource Index
| File | Purpose |
|---|---|
UnitProfileSchema.md |
Data model - 7 dimensions with typed fields |
QueryTemplates.md |
Parameterized search queries per dimension |
HigherEdContext.md |
Domain knowledge (Carnegie, org patterns, sources) |
UnitResearchLens.md |
Synthesis rules, confidence calibration, anti-hallucination |
UnitProfileTemplate.md |
Individual unit output template |
LeaderProfileTemplate.md |
Unit director profile template |
ComparativeReportTemplate.md |
Multi-unit comparison template |
ResearchBriefTemplate.md |
Executive summary for sharing |
Output Configuration
Default output directory: Current working directory, or user-specified path.
File naming:
- Single profile:
{institution-slug}-{unit-slug}-profile.md - Leader profile:
{director-name-slug}-leader-profile.md - Comparative report:
{category-slug}-comparative-report.md - Research brief:
{category-slug}-research-brief.md
Offer to save after generating any output. If the user specifies a directory up front, save there automatically.
Usage Examples
Single unit, standard depth:
"Profile the Center for Teaching and Learning at Stanford"
Single unit, quick:
"Who runs the digital learning office at Georgia Tech?"
Multi-unit comparison:
"Research AI centers at R1 universities - find 10 and compare them"
Multi-unit, scoped:
"Compare teaching and learning centers across Big Ten schools, deep dive"
Targeted update:
"Refresh the leadership section of the MIT Open Learning profile - they have a new director"
Quality Standards
All outputs must follow the rules in UnitResearchLens.md:
- No fabrication. Unknown fields are marked "Unknown" or "Not identified," never invented.
- Source attribution. Every section includes the URLs consulted.
- Confidence scoring. Staffing and budget estimates include confidence levels (0.0-1.0).
- Conflict documentation. When sources disagree, both are noted with resolution reasoning.
- Gap transparency. Missing data is explicitly listed, never silently omitted.