/report — Claude Code Skill

A Claude Code skill that generates beautiful, self-contained HTML reports instead of dumping long output into your terminal.

When Claude's response would exceed ~10 lines — research findings, audit results, task summaries, comparisons — it automatically generates a styled HTML report and opens it in your browser.

What it does

• Auto-triggers when output would be long (>10 lines) or when you type /report
• 5 report types: task-completion, research, audit, error-blocker, comparison
• Dark-themed with PicoCSS — clean, readable, professional
• Indexed — every report gets JSON metadata for search and dashboards
• Self-contained — each report is a single HTML file, no build step
• Zero dependencies — Python stdlib only

Example

Instead of a wall of terminal text, you get this:

See the full HTML: examples/example-task-completion.html (download and open in browser).

Report: Security Audit Results → ~/claude-reports/myproject/2026-04-10/143022-security-audit-results.html

The report opens in your browser with structured sections, pass/fail tables, color-coded badges, and collapsible details.

Install

1. Add the skill to Claude Code

Copy the SKILL.md file to your Claude Code skills directory:

mkdir -p ~/.claude/skills/report
cp SKILL.md ~/.claude/skills/report/

2. Install the CLI tool

The bin/report script handles file paths, indexing, and opening reports. Add it to your PATH:

cp bin/report ~/.local/bin/report
chmod +x ~/.local/bin/report

Or symlink it:

ln -s "$(pwd)/bin/report" ~/.local/bin/report

Make sure ~/.local/bin is in your PATH.

3. Verify

report --help

Usage

Just use Claude Code normally. The skill triggers automatically when appropriate.

You can also explicitly request a report:

/report

Or ask Claude to generate a specific type:

"Audit this codebase for security issues"     → audit report
"Compare React vs Svelte for this project"    → comparison report
"What did you change?"                         → task-completion report
"Research the best CI/CD options"              → research report

Report types

Type	Sections	When to use
task-completion	Summary, Changes Made, Verification, Next Steps	After implementing something
research	Key Findings, Analysis, Recommendations, Sources	Research and investigation
audit	Score, Pass/Fail Table, Findings, Recommendations	Quality/security/perf checks
error-blocker	What's Blocked, Root Cause, Action Items, Workarounds	When something is broken
comparison	Options Table, Detailed Analysis, Recommendation	Evaluating alternatives

Where reports go

~/claude-reports/
  ├── index.json              # Auto-maintained index of all reports
  ├── myproject/
  │   └── 2026-04-10/
  │       ├── 143022-security-audit.html
  │       └── 150511-api-research.html
  └── another-project/
      └── ...

CLI reference

# Generate a file path for a new report
report --path --project "myproject" --title "Security Audit" --type audit

# Open a report in browser (also indexes it)
report --open ~/claude-reports/myproject/2026-04-10/143022-security-audit.html

# List recent reports (JSON)
report --list
report --list --project myproject --type audit --days 7

# Rebuild the index from disk
report --index

Platform support

• macOS: Uses open to launch browser
• Linux: Uses xdg-open to launch browser
• Windows: Uses os.startfile to launch browser

License

MIT