Skip to main content
batch operations allow you to create or delete multiple records concurrently, dramatically improving performance for bulk operations.

​
batch delete

delete multiple records in one command with automatic concurrency control.

​
multiple URIs as arguments

pdsx rm app.bsky.feed.post/abc123 app.bsky.feed.post/def456 app.bsky.feed.post/ghi789
output: 
✓ successfully deleted 3/3 records (100%)

​
from file via stdin

# one URI per line
cat uris.txt | pdsx rm
where uris.txt contains: 
app.bsky.feed.post/abc123
app.bsky.feed.post/def456
app.bsky.feed.post/ghi789

​
control concurrency

# default is 10 concurrent operations
pdsx rm uri1 uri2 uri3 --concurrency 20

# fail on first error instead of continuing
pdsx rm uri1 uri2 uri3 --fail-fast

​
batch create

create multiple records from JSONL (JSON Lines) format.

​
single record (backward compatible)

pdsx create app.bsky.feed.post text="hello world"

​
batch create from JSONL

cat posts.jsonl | pdsx create app.bsky.feed.post
where posts.jsonl contains one JSON object per line: 
{"text": "first post"}
{"text": "second post"}
{"text": "third post", "langs": ["en"]}
output: 
✓ successfully created 3/3 records (100%)

​
control concurrency

cat posts.jsonl | pdsx create app.bsky.feed.post --concurrency 20
cat posts.jsonl | pdsx create app.bsky.feed.post --fail-fast

​
batch update

update multiple records from JSONL format with uri field.

​
single update (backward compatible)

pdsx update app.bsky.feed.post/abc123 text="updated text"

​
batch update from JSONL

cat updates.jsonl | pdsx update
where updates.jsonl contains one JSON object per line with a uri field: 
{"uri": "app.bsky.feed.post/abc123", "text": "new text for post 1"}
{"uri": "app.bsky.feed.post/def456", "text": "new text for post 2", "langs": ["en"]}
{"uri": "app.bsky.feed.post/ghi789", "text": "new text for post 3"}
output: 
✓ successfully updated 3/3 records (100%)
note: the uri field can be a shorthand URI (just the record key) or a full AT-URI. the uri field is extracted from each JSON object, and the remaining fields are used as the update data.

​
control concurrency

cat updates.jsonl | pdsx update --concurrency 20
cat updates.jsonl | pdsx update --fail-fast

​
JSONL format

JSONL (JSON Lines) is one JSON object per line: 
{"field1": "value1", "field2": "value2"}
{"field1": "value3", "field2": "value4"}
{"field1": "value5"}
rules:
  • one complete JSON object per line
  • no commas between objects
  • empty lines are ignored
  • no array wrapper [...]
generate JSONL: 
# from jq
pdsx ls app.bsky.feed.post -o json | jq -c '.[]' > posts.jsonl

# manually
echo '{"text": "post 1"}' > posts.jsonl
echo '{"text": "post 2"}' >> posts.jsonl

​
performance

batch operations use concurrent execution with rate limiting:
  • default concurrency: 10 simultaneous operations
  • respects rate limits: semaphore-based throttling
  • progress tracking: shows progress bar in interactive terminals
  • error handling: continues on error by default, reports failures at end

​
real-world performance

sequential deletion (old way): 
for uri in $(cat uris.txt); do pdsx rm $uri; done
# 82 records: ~41 seconds (0.5s per record)
batch deletion (new way): 
cat uris.txt | pdsx rm
# 82 records: ~8 seconds (10 concurrent)
5x faster for bulk operations.

​
common workflows

​
backup and restore

# backup all posts to JSONL
pdsx ls app.bsky.feed.post --limit 1000 -o json | jq -c '.[]' > backup.jsonl

# restore from backup (be careful!)
cat backup.jsonl | pdsx create app.bsky.feed.post

​
bulk delete by pattern

# find test posts and delete them
pdsx ls app.bsky.feed.post --limit 100 -o json \
  | jq -r '.[] | select(.text | contains("[test]")) | .uri' \
  | sed 's|at://did:plc:[^/]*/||' \
  | pdsx rm

​
migrate records

# export from one collection
pdsx ls app.custom.collection -o json | jq -c '.[]' > export.jsonl

# import to another (after editing JSONL)
cat export.jsonl | pdsx create app.custom.newcollection

​
bulk update posts

# find posts to update and generate JSONL with modifications
pdsx ls app.bsky.feed.post --limit 100 -o json \
  | jq -c '.[] | select(.text | contains("[draft]")) | {uri: (.uri | sub("at://[^/]+/"; "")), text: (.text | sub("\\[draft\\] "; ""))}' \
  | pdsx update

​
limitations

  • authentication required: batch operations need valid credentials
  • rate limits apply: respect PDS rate limits with --concurrency
  • memory usage: large batches held in memory during processing

​
error handling

​
continue on error (default)

cat uris.txt | pdsx rm
if some URIs fail, it continues and reports: 
✓ successfully deleted 80/82 records (97%)

✗ 2 operations failed:
  abc123: Record not found
  def456: Invalid URI format

​
fail-fast mode

cat uris.txt | pdsx rm --fail-fast
stops immediately on first error - useful for testing.