export

name	export
description	export the lore knowledge graph or its slices to json, csv, or markdown
tools	Bash, Read, AskUserQuestion

When the user runs /lore:export, write a portable snapshot of the lore data they care about. Default destination is the current working directory; respect any path the user gives.

locate the database

DB="$HOME/.claude/lore/lore.db"
[ -f "$DB" ] || DB="$HOME/.claude/mine.db"
if [ ! -f "$DB" ]; then echo "no lore.db -- run /lore once to seed it"; exit 0; fi

intents

user phrasing	export	format
"export notes", "save my notes"	notes table	json (default) or markdown
"export sessions"	sessions table	json or csv
"export project X"	project slice (sessions + notes + file list)	json bundle
"export cooccurrences"	file_cooccurrences view	csv (large -- prefer csv)
"export everything", "full snapshot"	all tables	json bundle in a directory

If the user doesn't specify a format, infer: bulk graph data → csv (smaller), structured/nested → json, anything they'll read by eye → markdown.

destination

Default: ./lore-export-<YYYYMMDD>.<ext> in cwd. If the user gives a path, use it. If a file already exists at the destination, ask before overwriting.

queries

notes (json)

OUT="${1:-./lore-notes-$(date +%Y%m%d).json}"
sqlite3 -json "$DB" "SELECT * FROM notes ORDER BY created_at DESC;" > "$OUT"
echo "wrote $OUT ($(wc -l < "$OUT") lines)"

notes (markdown)

OUT="${1:-./lore-notes-$(date +%Y%m%d).md}"
{
  echo "# lore notes export — $(date -u +%Y-%m-%dT%H:%M:%SZ)"
  echo
  sqlite3 "$DB" "SELECT id, note_type, title, project_name, tags, created_at, body FROM notes ORDER BY created_at DESC;" |
    awk -F'|' '{
      printf "## #%s [%s] %s\n", $1, $2, $3;
      if ($4 != "") printf "- project: %s\n", $4;
      if ($5 != "") printf "- tags: %s\n", $5;
      printf "- created: %s\n\n", $6;
      if ($7 != "") printf "%s\n\n", $7;
    }'
} > "$OUT"
echo "wrote $OUT"

sessions (csv)

OUT="${1:-./lore-sessions-$(date +%Y%m%d).csv}"
sqlite3 -header -csv "$DB" "SELECT id, project_name, model, start_time, duration_active_seconds, total_input_tokens, total_output_tokens FROM sessions WHERE is_subagent = 0 ORDER BY start_time DESC;" > "$OUT"
echo "wrote $OUT"

project bundle (json)

PROJECT="$1"
OUT="${2:-./lore-$PROJECT-$(date +%Y%m%d).json}"
python3 <<PY > "$OUT"
import json, sqlite3
db = sqlite3.connect("$DB")
db.row_factory = sqlite3.Row
project = "$PROJECT"
def rows(sql, *args):
    return [dict(r) for r in db.execute(sql, args).fetchall()]
bundle = {
  "project": project,
  "exported_at": __import__("datetime").datetime.utcnow().isoformat() + "Z",
  "sessions": rows("SELECT * FROM sessions WHERE project_name = ? AND is_subagent = 0 ORDER BY start_time DESC", project),
  "notes": rows("SELECT * FROM notes WHERE project_name = ? ORDER BY created_at DESC", project),
  "files": rows(
    "SELECT input_summary AS path, COUNT(DISTINCT session_id) AS sessions FROM tool_calls tc JOIN sessions s ON tc.session_id = s.id "
    "WHERE s.project_name = ? AND tc.tool_name IN ('Edit','Read','Write','MultiEdit') AND tc.input_summary IS NOT NULL "
    "GROUP BY input_summary ORDER BY sessions DESC", project),
}
print(json.dumps(bundle, indent=2, default=str))
PY
echo "wrote $OUT"

cooccurrences (csv)

OUT="${1:-./lore-cooccurrences-$(date +%Y%m%d).csv}"
echo "this view recomputes -- the export may take ~5s on a populated db"
sqlite3 -header -csv "$DB" "SELECT * FROM file_cooccurrences ORDER BY session_count DESC;" > "$OUT"
echo "wrote $OUT ($(wc -l < "$OUT") rows)"

full snapshot (directory of json files)

DIR="${1:-./lore-snapshot-$(date +%Y%m%d)}"
mkdir -p "$DIR"
for table in sessions messages tool_calls errors notes; do
  sqlite3 -json "$DB" "SELECT * FROM $table" > "$DIR/$table.json"
  echo "  $table.json: $(wc -c < "$DIR/$table.json") bytes"
done
sqlite3 -json "$DB" "SELECT * FROM file_cooccurrences" > "$DIR/file_cooccurrences.json"
echo "wrote $DIR/"

A full snapshot can be large -- warn the user before writing if messages would exceed ~50MB (rough check: sqlite3 ... "SELECT COUNT(*) FROM messages;" -- bail if >100k unless they confirm).

confirming destructive overwrites

If the destination file or directory already exists, list what would be overwritten and ask before clobbering. Never silently overwrite a non-empty export directory.

output

After a successful write, print the path and size in one line. If the user asked for a format the schema can't satisfy (e.g. "export the graph as graphml"), explain what's available rather than inventing an exporter.