mbwsims/claude-universe/hotspots

Hotspots

Find the riskiest files in the codebase — the ones where bugs are most likely to hide. Risk correlates with: high change frequency (churn), high complexity, many dependencies, and multiple authors.

A file that changes every week, has 500 lines of complex logic, and is imported by 20 other files is a ticking time bomb. This skill finds those files.

Workflow

0. Deterministic Analysis (if available)

If lenskit_analyze is available, call it without a file argument to get project-wide metrics — churn, complexity, coupling, and risk scores for all source files. Use these as the foundation for hotspot ranking, supplementing with qualitative assessment.

If unavailable, perform full manual analysis as described below.

0.5. Quick Health Probe

If lenskit_status is available, call it with detailed: true FIRST before the detailed analysis. With detailed: true it returns:

• Total file count and average risk score (calibration: is this a large/complex project?)
• Top risk files with scores (you may already have your answer)
• Circular dependency count (structural issue to note)
• Hub count (coupling signal)
• Test coverage ratio (context for risk assessment)

If the status result already provides a clear answer (e.g., obvious top risk files with high scores), you may skip the manual git analysis and jump to presenting findings.

1. Measure Churn

Use git history to find files that change most frequently:

git log --format=format: --name-only --since="6 months ago" | sort | uniq -c | sort -rn | head -30

This gives the top 30 most-changed files in the last 6 months. High churn = high risk.

Filtering churn results:

• Exclude lock files and generated code: grep -v -E '(package-lock|yarn.lock|pnpm-lock|\.generated\.|\.min\.)'
• Focus on source files only: grep -E '\.(ts|tsx|js|jsx|py|go|rs|java)$'
• For monorepos, scope to a specific package: add -- packages/my-package/ to the git log command

Also check for files with many different authors (indicates shared/critical code):

git log --format='%an' --since="6 months ago" -- {file} | sort -u | wc -l

2. Measure Complexity

For each high-churn file, assess complexity:

• Line count: Files over 300 lines are harder to reason about
• Function count: Files with many exported functions have broad surface area
• Nesting depth: Deeply nested conditionals indicate complex logic
• Cyclomatic complexity: Count decision points (if, switch, ternary, catch, &&, ||)

Read the file and make a qualitative assessment if quantitative tools aren't available.

3. Measure Coupling

For each high-churn file, check how many other files depend on it:

grep -rl "from.*{module-name}" src/ | wc -l

High import count = wide blast radius. Changes to this file affect many others.

4. Score and Rank

Combine the signals into a risk score:

Rank files by combined risk. The top 5-10 are the hotspots.

Amplify with graph data: If lenskit_graph is available, check whether any hotspot is also a hub file (high importer count) or part of a circular dependency. Hub status amplifies risk — a complex, high-churn file that 15 other files depend on is a higher priority than one with zero importers.

5. Present Findings

Report format:

## Hotspots — {project}
 
Analyzed {n} files over {timeframe}
 
### Top Hotspots
 
| Rank | File | Churn | Size | Dependents | Risk |
|------|------|-------|------|------------|------|
| 1 | src/lib/auth.ts | 24 changes | 342 lines | 18 importers | Critical |
| 2 | src/app/api/orders/route.ts | 19 changes | 287 lines | 3 importers | High |
| 3 | ... |
 
### Hotspot Details
 
**#1: src/lib/auth.ts** — Critical Risk
- Changed 24 times in 6 months by 4 different authors
- 342 lines with 12 exported functions
- Imported by 18 other files
- Suggestion: This file does too much. Split auth verification,
  session management, and token handling into separate modules.
 
**#2: src/app/api/orders/route.ts** — High Risk
- ...
 
### Cool Spots (Low Risk, Stable)
{Note 2-3 files that are stable, simple, and well-isolated — positive examples}

Monorepo Guidance

For monorepos with multiple packages:

• Run analysis scoped to each package separately: lenskit_analyze on each package root
• Compare hotspot scores ACROSS packages to find the worst areas
• Note cross-package dependencies (a hotspot in a shared package affects all consumers)
• If a shared package has high churn, it's a higher-priority hotspot than a leaf package

with the same churn score

Related Skills

• `/impact` — Check blast radius before touching a hotspot
• `/map` — Inspect surrounding layers before proposing structural changes

Guidelines

• Focus on source files, not config or generated files. Filter out node_modules/,

package-lock.json, migration files, etc.

• A file that changes often but is small and simple (e.g., constants.ts) is not a

hotspot. Churn alone is not risk — churn x complexity is.

• Include "cool spots" — stable, well-structured files that are positive examples.

This makes the report balanced and gives the team good patterns to follow.

• If git history is short (new project), note this and focus on complexity/coupling instead.