Intelligence reports

The Intelligence page generates and delivers a weekly bug-intelligence digest per project. Think of it as the standup your bug tracker never gave you — written in plain English, backed by a week of real data, delivered to your inbox and available on demand in the console.

What’s in a report

Each digest covers the previous calendar week and contains six sections:

Section	What it shows
Executive summary	3–5 sentence narrative generated by the LLM from all the data below
Top 10 components by volume	Report count per component + week-over-week delta (↑/↓ with %)
Emerging Bug-Ontology tags	Tags that appeared or spiked this week vs. the trailing 4-week baseline
Critical-bug backlog	Top 5 longest-open critical bugs with age, component, and last-touched date
Fix velocity	Fix attempts opened / merged / failed this week; median time from report to PR
Rewards snapshot	New contributors, points issued, tier-ups (if rewards is enabled)

Schedule and delivery

Reports are generated by the intelligence-report edge function on a Monday 04:00 UTC cron. The generated HTML is persisted in project_settings.intelligence_cache and the PDF is uploaded to your configured BYO Storage bucket (or Mushi-managed storage on Cloud plans).

Delivery channels:

Channel	How to configure
In-console	Always available — click any week in the calendar to read the report
Email digest	Settings → Intelligence → Email recipients (accepts multiple addresses)
Slack	Settings → Integrations → Slack → enable Weekly digest
Webhook	`intelligence.generated` event fires with the PDF URL in the payload

Regenerating on demand

Click Regenerate on any week to re-run the intelligence-report function against that week’s data. Useful when:

You’ve just triaged a backlog and want the narrative to reflect the new state.
The scheduled run failed (visible as an error state on the week card).
You want to preview the report before Monday.

Regeneration overwrites the cached HTML and PDF for that week. The previous version is not archived.

Regeneration is rate-limited to 3 times per week per project on Indie plans and unlimited on Pro and Enterprise.

Cross-customer benchmarks

The optional Benchmarking section shows how your project’s report volume, severity mix, and fix velocity compare against the anonymised aggregate of other projects in the same industry vertical.

Benchmarks are opt-in:

Toggle Benchmarking in Settings → Intelligence.
Select your industry vertical (SaaS / Mobile / E-commerce / Fintech / Other).
Confirm that you consent to contributing your project’s anonymised aggregate metrics to the benchmark pool.

Mushi enforces k ≥ 5 anonymity: a benchmark data point is only included in the digest when at least 5 other projects contribute to it, so no individual project’s data is identifiable.

PDF format

The PDF renders from the persisted rendered_html — no re-fetch or re-generation required at download time. PDFs are stored in:

Mushi Cloud: your project’s isolated storage prefix in the shared S3 bucket.
Self-hosted / BYO Storage: the intelligence-reports/ prefix in your own bucket. Configure the bucket in Settings → Storage.

PDF URLs are signed (15-minute expiry) and require the viewer to be authenticated to the project. They are not publicly accessible.