refactor: callback-based processor design #294

andreatgretel · 2026-02-04T02:23:16Z

Summary

Refactors the processor system from stage-based configuration to a callback-based design. Processors define behavior by implementing callback methods instead of setting a build_stage field in config. The preprocessor stage has been removed in favor of the simpler pre-batch approach.

Changes

Added

Three callback methods on Processor base class with default no-op implementations:
- process_before_batch() - after seed columns, before dependent columns
- process_after_batch() - after each batch completes
- process_after_generation() - on final combined dataset after all batches
implements() method to check if a processor overrides a callback
ProcessorRunner class encapsulating all processor execution logic with public processors property
replace_buffer() on DatasetBatchManager for PRE_BATCH processors that change row count
Row-count-change logging in _run_stage fires for all stages automatically
Comprehensive test suite: stage ordering, error handling, no-op processors, definition order, batch partitioning (even/uneven/fewer-than-batch-size), empty DataFrames

Changed

Extracted processor execution from ColumnWiseBuilder into dedicated ProcessorRunner
Both pre-batch and post-batch processing now run inside _run_batch, keeping the full batch lifecycle in one place
run_after_generation(batch_size) re-chunks output by batch_size to avoid empty parquet files when processors reduce row count
ProcessorRunner only depends on ArtifactStorage (not ResourceProvider)
Updated docs/concepts/processors.md with 3-stage callback design

Removed

preprocess() stage and all associated plumbing (preprocessed_seed_uri, cleanup_preprocessed_seed, disk I/O for seed preprocessing)
BuildStage enum and fixed wrong return type on get_processor_configs()
build_stage field from ProcessorConfig

Attention Areas

Reviewers: Please pay special attention to the following:

processor_runner.py - Core processor execution logic with batch-size-based chunking in run_after_generation
column_wise_builder.py#_run_batch - Full batch lifecycle: generate → post-batch process → write → finish
dataset_batch_manager.py - replace_buffer() for PRE_BATCH row-count changes (note: overlaps with update_records(allow_resize=True) in feat: add allow_resize for 1:N and N:1 generation patterns #286 - should unify when both land)

Description updated with AI

andreatgretel · 2026-02-05T15:50:37Z

packages/data-designer-config/src/data_designer/config/processors.py

    name: str = Field(
        description="The name of the processor, used to identify the processor in the results and to write the artifacts to disk.",
    )
-    build_stage: BuildStage = Field(


No need for build_stage anymore

Sweet, this is a neat approach!

andreatgretel · 2026-02-05T15:51:20Z

packages/data-designer-engine/src/data_designer/engine/processing/processors/base.py

 from data_designer.engine.configurable_task import ConfigurableTask, DataT, TaskConfigT


 class Processor(ConfigurableTask[TaskConfigT], ABC):


This is the main change - three different methods

greptile-apps · 2026-02-05T15:54:03Z

Greptile Overview

Greptile Summary

Refactors processor system from config-based stages to callback-based design. Processors now define behavior by implementing process_before_batch(), process_after_batch(), or process_after_generation() instead of setting a build_stage field. The refactor removes the preprocessor stage in favor of the simpler pre-batch approach and extracts all processor execution into a dedicated ProcessorRunner class.

Key changes:

New ProcessorRunner encapsulates execution logic with automatic row-count-change logging across all stages
Three callback methods on Processor base class with no-op defaults; implements() detects which callbacks are overridden
Pre-batch runs after seed generation but before dependent columns; post-batch runs after each batch; after-generation runs once on final combined dataset
DatasetBatchManager.replace_buffer() added to support PRE_BATCH processors that change row count
run_after_generation() re-chunks output by batch_size to maintain consistent file sizes when processors add/remove rows
Removed build_stage config field and BuildStage enum entirely
Comprehensive test coverage for stage ordering, batch partitioning edge cases, and empty DataFrames

Issues found:

When an AFTER_GENERATION processor filters dataset to 0 rows, the re-chunking logic writes one empty parquet file due to max(len(df), 1) forcing one iteration

Confidence Score: 4/5

This PR is safe to merge with one minor edge case to address
Clean architectural refactor with comprehensive tests. One edge case identified: empty dataset after AFTER_GENERATION processing writes one empty parquet file. Impact is low since it only affects builds where a processor completely filters out all rows, which is rare. The callback-based design is well-structured, test coverage is thorough, and the migration from config-based to callback-based stages is complete.
packages/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py - empty dataset edge case in run_after_generation()

Important Files Changed

Filename	Overview
packages/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py	New file implementing callback-based processor execution. Clean separation of concerns with stage enum. Row-count change logging handles all stages automatically. Empty DataFrame edge case writes one empty file intentionally for consistency.
packages/data-designer-engine/src/data_designer/engine/processing/processors/base.py	Clean callback-based design with three no-op default methods. `implements()` uses method resolution to detect overrides correctly.
packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py	Refactored to use `ProcessorRunner`. Pre-batch runs after seed generation, post-batch runs in `_run_batch()`, after-generation runs in `build()`. Removed stage-based dict, now uses simple processor list.
packages/data-designer-engine/src/data_designer/engine/dataset_builders/utils/dataset_batch_manager.py	Added `replace_buffer()` method to support PRE_BATCH processors that change row count. Updates both buffer and batch size tracking.

Sequence Diagram

sequenceDiagram
    participant Builder as ColumnWiseBuilder
    participant Runner as ProcessorRunner
    participant BatchMgr as DatasetBatchManager
    participant Processor as Processor(s)
    participant Storage as ArtifactStorage

    Note over Builder: build(num_records)
    Builder->>Builder: Initialize generators
    Builder->>BatchMgr: start(num_records, buffer_size)
    
    loop For each batch
        Builder->>Builder: _run_batch()
        Note over Builder: Generate seed columns
        Builder->>Runner: run_pre_batch(batch_manager)
        Runner->>BatchMgr: get_current_batch(as_dataframe=True)
        BatchMgr-->>Runner: DataFrame
        Runner->>Processor: process_before_batch(df)
        Processor-->>Runner: Modified DataFrame
        Runner->>BatchMgr: replace_buffer(records)
        Note over Builder: Generate dependent columns
        Builder->>Runner: run_post_batch(df, batch_number)
        Runner->>Processor: process_after_batch(df, batch_number)
        Processor-->>Runner: Modified DataFrame
        Runner-->>Builder: Processed DataFrame
        Builder->>BatchMgr: update_records(processed_df)
        Builder->>BatchMgr: write() + finish_batch()
    end
    
    Builder->>BatchMgr: finish()
    Builder->>Runner: run_after_generation(batch_size)
    Runner->>Storage: load_dataset()
    Storage-->>Runner: Complete DataFrame
    Runner->>Processor: process_after_generation(df)
    Processor-->>Runner: Final DataFrame
    Runner->>Storage: rmtree(final_dataset_path)
    loop Re-chunk by batch_size
        Runner->>Storage: write_batch_to_parquet_file()
    end

greptile-apps

_{6 files reviewed, 3 comments}

_{Edit Code Review Agent Settings | Greptile}

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py

...s/data-designer-engine/src/data_designer/engine/column_generators/generators/seed_dataset.py

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py

greptile-apps

_{7 files reviewed, 4 comments}

_{Edit Code Review Agent Settings | Greptile}

...ages/data-designer-engine/src/data_designer/engine/processing/processors/schema_transform.py

greptile-apps · 2026-02-05T16:16:52Z

Additional Comments (3)

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py
Leaking DuckDB connection
_run_pre_generation_processors() creates a DuckDB connection via seed_reader.create_duckdb_connection() and never closes it. If a build runs many times in-process (service mode/tests), this will leak connections/file handles. Close the connection (e.g., try/finally: conn.close()), or use a context manager if the seed reader supports it.

Prompt To Fix With AI

This is a comment left during a code review.
Path: packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py
Line: 438:444

Comment:
**Leaking DuckDB connection**
`_run_pre_generation_processors()` creates a DuckDB connection via `seed_reader.create_duckdb_connection()` and never closes it. If a build runs many times in-process (service mode/tests), this will leak connections/file handles. Close the connection (e.g., `try/finally: conn.close()`), or use a context manager if the seed reader supports it.

How can I resolve this? If you propose a fix, please make it concise.

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py
seed_row_indices not updated
_run_pre_generation_processors() only sets resource_provider.seed_row_indices when len(df) != original_len. If preprocess reorders rows or filters+adds back to the same length, seed_row_indices stays unset and the generator will read unfiltered rows even though preprocess changed the seed selection. Track indices whenever preprocess runs (or whenever _dd_rowid changes), not only on length change.

Prompt To Fix With AI

This is a comment left during a code review.
Path: packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py
Line: 448:455

Comment:
**seed_row_indices not updated**
`_run_pre_generation_processors()` only sets `resource_provider.seed_row_indices` when `len(df) != original_len`. If preprocess reorders rows or filters+adds back to the same length, `seed_row_indices` stays unset and the generator will read unfiltered rows even though preprocess changed the seed selection. Track indices whenever preprocess runs (or whenever `_dd_rowid` changes), not only on length change.

How can I resolve this? If you propose a fix, please make it concise.

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py
Postprocess changes may be lost
_run_post_generation_processors() only rewrites the final dataset when row count or column names/order change. If postprocess() modifies values in-place (same rows/columns), those changes won't be persisted and will be silently dropped. This is a functional correctness issue for value-only postprocessors; consider always rewriting when processors exist, or require postprocessors to return a new df and detect df is not original_df/hashing.

Prompt To Fix With AI

This is a comment left during a code review.
Path: packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py
Line: 486:496

Comment:
**Postprocess changes may be lost**
`_run_post_generation_processors()` only rewrites the final dataset when row count or column *names/order* change. If `postprocess()` modifies values in-place (same rows/columns), those changes won't be persisted and will be silently dropped. This is a functional correctness issue for value-only postprocessors; consider always rewriting when processors exist, or require postprocessors to return a new df and detect `df is not original_df`/hashing.

How can I resolve this? If you propose a fix, please make it concise.

greptile-apps

_{6 files reviewed, 3 comments}

_{Edit Code Review Agent Settings | Greptile}

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py

...s/data-designer-engine/src/data_designer/engine/column_generators/generators/seed_dataset.py

tests_e2e/tests/test_health_checks.py

...s/data-designer-engine/src/data_designer/engine/column_generators/generators/seed_dataset.py

greptile-apps

_{6 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py

docs/concepts/processors.md

johnnygreco · 2026-02-06T16:54:10Z

docs/concepts/processors.md

+|-------|--------------|-----------------|-----------|
+| Pre-generation | Once, on full seed data before batching | `preprocess()` | Filter seed data, validate inputs, normalize data |
+| Pre-batch | After seed columns, before dependent columns | `process_before_batch()` | Transform seed data before other columns are generated |
+| Post-batch | After each batch completes | `process_after_batch()` | Drop columns, transform schema per batch |


Somehow in by head dropping columns needs to happen post generation, BUT the key point is that each batch contains the full schema of the dataset. Might be good to have a tip (or something) box that highlights this point.

added an info box to the docs. dropping columns post-batch is probably better than post-generation since we don't need to write dropped columns to disk.

packages/data-designer-config/src/data_designer/config/processors.py

...s/data-designer-engine/src/data_designer/engine/column_generators/generators/seed_dataset.py

- Move imports to top of file - Add seed_data_setup and builder_with_seed fixtures - Add create_mock_processor helper function - Add edge case tests for exceptions, no-op processors, ordering

- Move all processor stage logic to new ProcessorRunner in utils/ - ProcessorRunner takes dependencies and provides complete stage methods - Builder now calls runner methods directly instead of wrapper methods - Remove unused imports from column_wise_builder

- Add ProcessorStage enum to replace raw strings in processor_runner - Rename _processors property to public processors + set_processor_runner - Fix docstring to include process_before_batch callback - Add info box to processors docs about full schema in batches - Add clarifying comment about pre-batch processing location Co-authored-by: Cursor <cursoragent@cursor.com>

Co-authored-by: Cursor <cursoragent@cursor.com>

Add replace_records() to DatasetBatchManager that replaces the buffer without requiring matching length. Use it in run_pre_batch() so processors that filter or expand rows don't crash.

- Restore schema transform preview write so processor artifacts are available in preview mode (read back by data_designer.py) - Return tuple from processors property to prevent mutation

…eration - Remove preprocess() callback, run_preprocess(), cleanup_preprocessed_seed(), preprocessed_seed_uri, and all related plumbing - Rename postprocess() to process_after_generation() across base class, ProcessorRunner, builder, config enums, docs, and tests - Drop resource_provider parameter from ProcessorRunner (no longer needed)

- Remove dead BuildStage enum and fix wrong return type on get_processor_configs() - Log row count delta instead of absolute values in PRE_BATCH - Preserve batch partitioning in run_after_generation() - Consolidate and simplify processor tests

andreatgretel · 2026-02-11T13:28:21Z

@johnnygreco @nabinchha removed preprocessor following our discussion offline

greptile-apps

_{17 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py

Avoids empty parquet files when processors reduce row count below the original file count. Re-chunks by buffer_size instead of matching the number of input files.

greptile-apps

_{17 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py

...ata-designer-engine/src/data_designer/engine/dataset_builders/utils/dataset_batch_manager.py

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py

- Rename replace_records to replace_buffer - Move row-count-change logging into _run_stage for all stages - Expose processors as public property on ProcessorRunner

Both pre-batch and post-batch now run inside _run_batch, keeping the full batch lifecycle (generate, process, write, finish) in one place. Preview mode skips post-batch via current_batch_number=None guard.

…ocessor-plugins # Conflicts: # packages/data-designer-engine/tests/engine/dataset_builders/test_column_wise_builder.py

greptile-apps

_{17 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py

greptile-apps

_{17 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

greptile-apps · 2026-02-12T00:34:20Z

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py

+        df = self._run_stage(df, ProcessorStage.AFTER_GENERATION)
+
+        shutil.rmtree(self._artifact_storage.final_dataset_path)
+        for i in range(0, max(len(df), 1), batch_size):


when AFTER_GENERATION processor filters dataset to 0 rows, max(len(df), 1) forces one iteration → writes one empty parquet file. consider early return if len(df) == 0 to avoid empty output files.

Prompt To Fix With AI

This is a comment left during a code review. Path: packages/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py Line: 98:98 Comment: when AFTER_GENERATION processor filters dataset to 0 rows, `max(len(df), 1)` forces one iteration → writes one empty parquet file. consider early return if `len(df) == 0` to avoid empty output files. How can I resolve this? If you propose a fix, please make it concise.

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch 2 times, most recently from 35df5cf to ec54289 Compare February 4, 2026 23:06

andreatgretel changed the title ~~feat: add processor plugins with PRE_GENERATION and POST_GENERATION stages~~ refactor: replace stage-based config with callback-based processor design Feb 4, 2026

andreatgretel mentioned this pull request Feb 4, 2026

feat: add processor plugin system #298

Closed

3 tasks

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch 2 times, most recently from edbaf44 to 5cc6f21 Compare February 5, 2026 14:05

andreatgretel mentioned this pull request Feb 5, 2026

feat: add processor plugin system #299

Draft

3 tasks

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch from f6a820b to 8310910 Compare February 5, 2026 15:48

andreatgretel marked this pull request as ready for review February 5, 2026 15:49

andreatgretel requested a review from a team as a code owner February 5, 2026 15:49

andreatgretel commented Feb 5, 2026

View reviewed changes

greptile-apps bot reviewed Feb 5, 2026

View reviewed changes

...ages/data-designer-engine/src/data_designer/engine/processing/processors/schema_transform.py Show resolved Hide resolved

greptile-apps bot reviewed Feb 5, 2026

View reviewed changes

johnnygreco reviewed Feb 5, 2026

View reviewed changes

tests_e2e/tests/test_health_checks.py Show resolved Hide resolved

johnnygreco reviewed Feb 5, 2026

View reviewed changes

...s/data-designer-engine/src/data_designer/engine/column_generators/generators/seed_dataset.py Outdated Show resolved Hide resolved

greptile-apps bot reviewed Feb 5, 2026

View reviewed changes

packages/data-designer-engine/src/data_designer/engine/dataset_builders/column_wise_builder.py Outdated Show resolved Hide resolved

johnnygreco reviewed Feb 5, 2026

View reviewed changes

docs/concepts/processors.md Outdated Show resolved Hide resolved

andreatgretel marked this pull request as draft February 5, 2026 17:47

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch 2 times, most recently from a1323f9 to 46880e7 Compare February 5, 2026 22:36

andreatgretel marked this pull request as ready for review February 5, 2026 22:38

andreatgretel requested review from mikeknep and nabinchha February 6, 2026 15:09

johnnygreco reviewed Feb 6, 2026

View reviewed changes

packages/data-designer-config/src/data_designer/config/processors.py Outdated Show resolved Hide resolved

johnnygreco reviewed Feb 6, 2026

View reviewed changes

...s/data-designer-engine/src/data_designer/engine/column_generators/generators/seed_dataset.py Outdated Show resolved Hide resolved

andreatgretel and others added 8 commits February 11, 2026 10:19

refactor processor tests with fixtures and add edge cases

4ea55c4

- Move imports to top of file - Add seed_data_setup and builder_with_seed fixtures - Add create_mock_processor helper function - Add edge case tests for exceptions, no-op processors, ordering

reword docs info box about schema availability during generation

5074981

Co-authored-by: Cursor <cursoragent@cursor.com>

support row-count changes in PRE_BATCH processors

56b29ad

Add replace_records() to DatasetBatchManager that replaces the buffer without requiring matching length. Use it in run_pre_batch() so processors that filter or expand rows don't crash.

fix preview artifact output and harden processors API

fec7af5

- Restore schema transform preview write so processor artifacts are available in preview mode (read back by data_designer.py) - Return tuple from processors property to prevent mutation

address PR review feedback

cd7031e

- Remove dead BuildStage enum and fix wrong return type on get_processor_configs() - Log row count delta instead of absolute values in PRE_BATCH - Preserve batch partitioning in run_after_generation() - Consolidate and simplify processor tests

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch from 21f8e8b to cd7031e Compare February 11, 2026 13:19

andreatgretel changed the title ~~refactor: replace stage-based config with callback-based processor design~~ refactor: callback-based processor design, remove preprocessor stage Feb 11, 2026

greptile-apps bot reviewed Feb 11, 2026

View reviewed changes

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py Outdated Show resolved Hide resolved

andreatgretel changed the title ~~refactor: callback-based processor design, remove preprocessor stage~~ refactor: callback-based processor design Feb 11, 2026

chunk after-generation output by batch_size

b6e4155

Avoids empty parquet files when processors reduce row count below the original file count. Re-chunks by buffer_size instead of matching the number of input files.

greptile-apps bot reviewed Feb 11, 2026

View reviewed changes

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py Show resolved Hide resolved

andreatgretel requested review from johnnygreco and nabinchha February 11, 2026 14:23

nabinchha reviewed Feb 11, 2026

View reviewed changes

andreatgretel added 2 commits February 11, 2026 17:02

address nabinchha review feedback

014ae30

- Rename replace_records to replace_buffer - Move row-count-change logging into _run_stage for all stages - Expose processors as public property on ProcessorRunner

move post-batch processing into _run_batch

a8715dc

Both pre-batch and post-batch now run inside _run_batch, keeping the full batch lifecycle (generate, process, write, finish) in one place. Preview mode skips post-batch via current_batch_number=None guard.

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch from a8715dc to 403bc69 Compare February 11, 2026 20:21

Merge remote-tracking branch 'origin/main' into andreatgretel/feat/pr…

a61848e

…ocessor-plugins # Conflicts: # packages/data-designer-engine/tests/engine/dataset_builders/test_column_wise_builder.py

andreatgretel force-pushed the andreatgretel/feat/processor-plugins branch from 403bc69 to a61848e Compare February 11, 2026 20:29

greptile-apps bot reviewed Feb 11, 2026

View reviewed changes

...ges/data-designer-engine/src/data_designer/engine/dataset_builders/utils/processor_runner.py Show resolved Hide resolved

nabinchha approved these changes Feb 11, 2026

View reviewed changes

Merge branch 'main' into andreatgretel/feat/processor-plugins

fe55c43

andreatgretel merged commit 429b558 into main Feb 12, 2026
47 checks passed

andreatgretel deleted the andreatgretel/feat/processor-plugins branch February 12, 2026 00:32

greptile-apps bot reviewed Feb 12, 2026

View reviewed changes

andreatgretel mentioned this pull request Feb 12, 2026

feat: add allow_resize for 1:N and N:1 generation patterns #286

Open

		from data_designer.engine.configurable_task import ConfigurableTask, DataT, TaskConfigT


		class Processor(ConfigurableTask[TaskConfigT], ABC):

refactor: callback-based processor design #294

refactor: callback-based processor design #294

Uh oh!

Conversation

andreatgretel commented Feb 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Added

Changed

Removed

Attention Areas

Uh oh!

andreatgretel Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

johnnygreco Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

andreatgretel Feb 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Feb 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Overview

Greptile Summary

Confidence Score: 4/5

Important Files Changed

Sequence Diagram

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

greptile-apps bot commented Feb 5, 2026

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

johnnygreco Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

andreatgretel Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

andreatgretel commented Feb 11, 2026

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

andreatgretel commented Feb 4, 2026 •

edited

Loading

andreatgretel Feb 5, 2026 •

edited

Loading

greptile-apps bot commented Feb 5, 2026 •

edited

Loading