Firefox History Search

Help the user find pages from their Firefox browsing history. The user will describe what they're looking for — a topic, a URL fragment, a time range, or a vague recollection. Your job is to query the history database and find it.

Setup

Firefox stores history in a SQLite database that is locked while Firefox is running. Always start by copying it:

cp ~/.mozilla/firefox/uq4slia4.default-release/places.sqlite /tmp/places_copy.sqlite

Database Schema

• moz_places: URLs and titles (id, url, title)
• moz_historyvisits: visit timestamps (place_id, visit_date)
• visit_date is in microseconds since epoch — divide by 1,000,000 for unix time

Query Patterns

Search by keyword (title or URL)

sqlite3 -header -column /tmp/places_copy.sqlite "
  SELECT datetime(v.visit_date/1000000,'unixepoch','localtime') as time,
         p.title, p.url
  FROM moz_historyvisits v JOIN moz_places p ON v.place_id=p.id
  WHERE p.title LIKE '%KEYWORD%' OR p.url LIKE '%KEYWORD%'
  ORDER BY v.visit_date DESC LIMIT 20;"

Show activity around a timestamp (context window)

Once you find an approximate time, show surrounding browsing activity (default ±30 minutes):

sqlite3 -header -column /tmp/places_copy.sqlite "
  SELECT datetime(v.visit_date/1000000,'unixepoch','localtime') as time,
         p.title, p.url
  FROM moz_historyvisits v JOIN moz_places p ON v.place_id=p.id
  WHERE datetime(v.visit_date/1000000,'unixepoch','localtime')
        BETWEEN 'YYYY-MM-DD HH:MM' AND 'YYYY-MM-DD HH:MM'
  ORDER BY v.visit_date;"

Browse a specific date

sqlite3 -header -column /tmp/places_copy.sqlite "
  SELECT datetime(v.visit_date/1000000,'unixepoch','localtime') as time,
         p.title, p.url
  FROM moz_historyvisits v JOIN moz_places p ON v.place_id=p.id
  WHERE date(v.visit_date/1000000,'unixepoch','localtime') = 'YYYY-MM-DD'
  ORDER BY v.visit_date;"

Workflow

1. Copy the database to /tmp
2. Based on the user's description, search by keyword, URL fragment, domain, or date
3. If the user is looking for something they found near another page, first find the anchor page, note its timestamp, then query a time window around it
4. Present results concisely — show time, title, and URL
5. If there are too many results, narrow with additional filters or tighter time windows
6. Clean up: rm /tmp/places_copy.sqlite when done

Tips

• Users often remember vaguely. Try multiple LIKE patterns if the first doesn't hit.
• Twitter/X links often appear as both twitter.com and x.com — search for both.
• t.co short URLs in history indicate a click-through from Twitter.
• The from_visit column in moz_historyvisits links to the referring page, useful for tracing navigation chains.

User's Request

$ARGUMENTS