fix(dynamodb, cosmosdb, mongodb): push filters down only where the source evaluates them as SQL does by lukekim · Pull Request #14419 · spiceai/spiceai

lukekim · 2026-09-26T16:26:49Z

📝 Summary

On trunk, filter pushdown returned wrong rows for MongoDB and DynamoDB and failed some valid queries, and Cosmos DB pushed nothing at all. Each connector now pushes a filter down as Exact only where the source selects exactly the rows SQL keeps from the rows the connector converts. Otherwise it pushes an Inexact superset that DataFusion filters again, and it pushes much more than before.

MongoDB (the translator lives in the fork, spiceai/datafusion-table-providers#77, which this pins):

$ne/$nin/$not/$nor match null and missing fields, so age <> 30 returned the rows with no age.
Comparisons are bracketed by BSON type, while the conversion renders ObjectIds, documents and symbols into strings.
Arrays match element-wise. NaN and -0.0 compare apart from Arrow.
A case-insensitive collection collation made s > 'Z' return nothing.
Casts were looked through where they change values.
The scan claimed MongoDB's sort order as the SQL one.
With unnesting, SELECT * failed with Path collision when a field is a document in some documents and a scalar in others.

DynamoDB:

NOT, <> and NOT IN are two-valued in DynamoDB, so they matched missing attributes.
Timestamp bounds were wrong for any offset and for layouts finer than the column.
A map key containing a dot was missed.
Attribute names with a space or -, and x IN (1, NULL), failed the query outright.
A partition-key equality or IN list now becomes one Query per partition. Sort-key predicates become the key condition, with the rest checked item by item.
ORDER BY the sort key sets ScanIndexForward, and a limit stops reading.
Filters beyond DynamoDB's 4 KB expression limit are applied to the rows read instead of failing the request.

Cosmos DB: projection, a parameterized WHERE (comparisons, IN, BETWEEN, null tests, LIKE 'prefix%', starts_with) and limits are pushed down. Each condition is guarded by the JSON type it compares, because the emulator and the service disagree on null and undefined. It also keeps a value its column cannot hold, so the query fails as it would unfiltered. An equality on the partition key therefore still reads every partition: a partition key of another JSON type lies in a logical partition of its own.

The starts_with guards translate only DataFusion's own function, never a user function registered under that name, and no other function is pushed.

Each connector gets a differential round-trip suite. Every query runs against the federated dataset and against an Arrow acceleration of the same source, which DataFusion evaluates locally, and the suite requires the same rows, compared cell by cell, plus that each case expected to push down did. The shared harness is crates/runtime/tests/pushdown_roundtrip/mod.rs.

🔗 Related

fix(mongodb): push down only filters MongoDB evaluates as SQL does, and stop claiming a sort order datafusion-table-providers#77, merged as 10b7fe71d7 on spiceai-54, which this pins. Its tree is that of c6cbc3a3ca, the branch head the evidence below was taken at.

📚 Docs

docs/dev/cosmosdb.md: a Push-down section, and the criteria table no longer says push-down is missing.
docs/dev/fork_patches.md: the new pin, and a row for the MongoDB patches with this PR's round-trip suite as their guard.

👀 Notes for Reviewers

Evidence. Every query ran against the federated dataset and against an Arrow acceleration of the same source, on mongo:8.0, amazon/dynamodb-local and the Cosmos DB Linux emulator (vnext-preview). Trunk is d62271b7a2.

Case	Trunk	This branch
MongoDB audit queries (25)	21 wrong	0 wrong, 20 pushed
MongoDB: symbol, collated collection and view, `TRY_CAST`, distant dates, dotted names with a catch-all	19 wrong	0 wrong
MongoDB: `unnest_depth: 1`, a field that is a document and a string	`Path collision at address.city`	rows
MongoDB: declared `timestamp` column, year-2262 date	read as 1677-09-21	NULL
DynamoDB audit queries (12)	5 wrong	0 wrong, all pushed
DynamoDB: attribute names with a space or `-`	`ExpressionAttributeNames contains invalid key`	rows
DynamoDB: `pk IN ('a', NULL)`	`Invalid FilterExpression: Syntax error`	right, a Query
DynamoDB: `-0.0` bounds, microsecond layout, `+15:00` offset, dotted map key, unpadded month	wrong	right
Cosmos DB	right rows, nothing pushed	right rows; `WHERE` and projection in the plan

For example, on trunk, over a collection collated {locale: "en", strength: 2}:

SELECT _id, s FROM coll WHERE s > 'Z'
federated:  []
truth:      [{"_id": 2, "s": "a"}, {"_id": 3, "s": "x"}, {"_id": 4, "s": "b"}]
plan:       MongoDBExec projection=[_id, s] filters=[{"s":{"$gt":"Z"}}]

On this branch the same query returns the truth rows, with collation=[simple] in the plan.

Adversarial reviews of this branch found more cases, each reproduced against the database before it was fixed. Among them: an Int64 sort key holding 9007199254740992.1 passing an IN check through f64 rounding; Cosmos DB's x = i64::MIN overflowing a guard into c["x"] > 9223372036854775807; seven 100-value IN lists reaching DynamoDB at 4,869 bytes after a second optimizer pass; NOT ((a OR b)), which DynamoDB rejects as redundant parentheses. Copilot's reviews found four more. The fork's filter that matches nothing, {_id: {$exists: false}}, matched every document of a view that projects _id away. It is {_id: {$in: []}} now, and the MongoDB suite queries such a view. Cosmos DB's single-partition read for pk = 'a' skipped a partition key of another JSON type, so the query returned rows where the unfiltered scan fails to decode that key. It is removed, and the Cosmos DB suite checks such a container. A DynamoDB condition on a flattened column like m.x.y checked only the nested path. It therefore dropped an item holding the column at two paths (m["x.y"] and m.x.y), which unnesting refuses. The condition now keeps an item with something at any other split of the name, and the DynamoDB suite checks such an item. Its projection also named only the first segment of a flattened column's name, so a column held under a top-level attribute such as m.x read as NULL; that predates this PR, and the projection now names every prefix of the name.

Behavior changes. These are bug fixes, but they change results for malformed or out-of-range data:

DynamoDB: a date or timestamp string that is not zero-padded as its layout is (2024- 1-01, 2024-1-02T…) now reads as NULL, as any other unparsable string does.
MongoDB: a date too distant for a declared microsecond or nanosecond column reads as NULL instead of wrapping.
MongoDB no longer claims a sort order, so ORDER BY … LIMIT sorts in DataFusion. Exact sort pushdown, through per-type queries merged in order, can follow.

Tests.

Unit tests: connector-dynamodb 220, connector-cosmosdb 45, and the fork's MongoDB module 164.
Integration: cargo test -p runtime --test integration --features mongodb,dynamodb,cosmosdb,duckdb -- --include-ignored pushdown_round_trips passes. That is 165 MongoDB cases (137 pushed), 117 DynamoDB (105 pushed) and 56 Cosmos DB (46 pushed). Over a container whose partition keys are of two JSON types, the Cosmos DB suite also requires a pushed-down equality on the partition key to fail on decoding, as the unfiltered scan does. The DynamoDB suite likewise requires a pushed-down condition to fail on an item that unnesting reads m.x.y from twice, and every item to read m.x.y from whichever attribute holds it.
The MongoDB and DynamoDB suites run in integration.yml. The Cosmos DB suite is #[ignore]d because the emulator is a large image, and CI doesn't enable cosmosdb anyway.
The DynamoDB integration module's EXPLAIN snapshot of a timestamp filter, dynamodb_timestamp_filter_pushdown, now shows the widened bounds: #n0 >= :v0 AND #n1 <= :v1 over 2024-11-30T12:34:56.123 and 2024-12-02T12:34:56.123~, where trunk compared 2024-12-01T17:34:56.123+05:00 as a string. The merge queue runs that module against a table in AWS and failed on the old snapshot. The new one is the plan the test prints against amazon/dynamodb-local holding the same items, which matched the queue's output byte for byte.
Clippy, with the lint gate's flags over these crates and the integration tests, is clean. The remote sign-off of 2a49ffa passed: Signed off by lukekim via remote — lint+test passed in 6922s. The remote sign-off of 9403797, that snapshot and a merge of trunk, is running.

Found along the way, not fixed here: SELECT truncate(1.5, 0) fails on trunk with Truncation width must be a positive Int64, got Float64. The round-trip suites use bucket to check that a Spice function stays out of the pushed filter.

github-actions · 2026-09-26T16:27:01Z

✅ Pull with Spice Passed

Passing checks:

✅ Title meets minimum length requirement (10 characters)
✅ No banned labels detected
✅ Has a label from required category kind/
✅ Has a label from required category area/
✅ Has at least one assignee: lukekim

Copilot

Copilot review overview

🟡 Changes recommended

DynamoDB and Cosmos DB can apply limits before retained inexact filters, causing valid matching rows to be omitted.

Review effort: Balanced
Findings: 2

Open (2)

Avoid limit pushdown with inexact Cosmos DB filters · New
Only push DynamoDB limits when all filters are exact · New

What changed in this PR

Improves SQL-correct filter pushdown for DynamoDB, Cosmos DB, and MongoDB.

Changes:

Adds semantics-aware DynamoDB and Cosmos DB filter/request planning.
Pins MongoDB pushdown fixes from the provider fork.
Adds differential integration tests and documentation.

File	Description
`Cargo.lock`	Updates locked dependencies.
`Cargo.toml`	Updates the table-providers fork pin.
`connector-cosmosdb/.../filter.rs`	Implements Cosmos DB filter translation.
`connector-cosmosdb/.../filter/tests.rs`	Tests Cosmos filter semantics.
`connector-cosmosdb/.../mod.rs`	Registers filter support.
`connector-cosmosdb/.../provider.rs`	Adds query, projection, partition, and limit pushdown.
`connector-dynamodb/src/arrow.rs`	Tightens date and timestamp parsing.
`connector-dynamodb/src/filter.rs`	Implements DynamoDB filter translation.
`connector-dynamodb/src/filter/tests.rs`	Tests translation edge cases.
`connector-dynamodb/src/lib.rs`	Registers the filter module.
`connector-dynamodb/src/provider.rs`	Integrates request, sort, and limit pushdown.
`connector-dynamodb/src/request_builder.rs`	Builds bounded Query and Scan requests.
`connector-dynamodb/src/request_builder/tests.rs`	Tests request planning.
`connector-dynamodb/src/request_plan.rs`	Expands request-plan metadata and variants.
`connector-dynamodb/src/table_schema.rs`	Adds key types and translation metadata.
`connector-dynamodb/src/utils.rs`	Removes the superseded filter visitor.
`crates/runtime/Cargo.toml`	Adds Cosmos test dependencies.
`runtime/tests/cosmosdb/mod.rs`	Registers the Cosmos round-trip suite.
`runtime/tests/cosmosdb/pushdown_roundtrip.rs`	Adds Cosmos differential tests.
`runtime/tests/dynamodb/mod.rs`	Registers the DynamoDB suite.
`runtime/tests/dynamodb/pushdown_roundtrip.rs`	Adds DynamoDB differential tests.
`runtime/tests/integration.rs`	Registers the shared harness.
`runtime/tests/mongo/mod.rs`	Registers the MongoDB suite.
`runtime/tests/mongo/pushdown_roundtrip.rs`	Adds MongoDB differential tests.
`runtime/tests/pushdown_roundtrip.rs`	Provides shared comparison utilities.
`docs/dev/cosmosdb.md`	Documents supported pushdown behavior.
`docs/dev/fork_patches.md`	Records the MongoDB fork patch.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

Copilot

Copilot review overview

🟡 Changes recommended

Inexact filters can interact incorrectly with pushed limits, and the differential harness can treat distinct results as equal.

Review effort: Balanced
Findings: 2 · 1

Open (3)

Only push DynamoDB limits when all filters are exact
Avoid limit pushdown with inexact Cosmos DB filters
Use structural row encoding to preserve distinct values · New

Copilot

Copilot review overview

🟡 Changes recommended

Cosmos single-partition routing can omit wrong-typed partition-key documents that its inexact filter intentionally retains.

Review effort: Balanced
Findings: 1

Open (1)

Unsafe single-partition routing for inexact equality filters · New

Resolved since last review (3)

Only push DynamoDB limits when all filters are exact
Avoid limit pushdown with inexact Cosmos DB filters
Use structural row encoding to preserve distinct values

cursor · 2026-09-26T23:13:49Z

Bugbot is paused — on-demand spend limit reached

Bugbot uses usage-based billing for this team and has hit its on-demand spend limit.

A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue.

Copilot

Copilot review overview

🟡 Changes recommended

DynamoDB pushdown can suppress a dotted-key collision error by filtering the ambiguous item before conversion.

Review effort: Balanced
Findings: 1

Open (1)

Widened filter drops rows with ambiguous flattened dotted keys · New

Resolved since last review (1)

Unsafe single-partition routing for inexact equality filters

Copilot

Copilot review overview

🔵 Needs a closer look

Core query-correctness logic spans a large connector change and an external temporary fork pin, preventing confident automated approval.

Review effort: Balanced
Findings: None

Resolved since last review (1)

Widened filter drops rows with ambiguous flattened dotted keys

lukekim · 2026-09-27T00:18:26Z

Addressed #14419 (review). That review had no line-comment thread. Its headline was that the query-correctness logic spans a large connector change and a temporary fork pin, which it leaves to a human reviewer, so I'm not requesting another Copilot pass.

The fork pin. spiceai/datafusion-table-providers#77 has to merge first. This PR then repoints Cargo.toml, Cargo.lock and docs/dev/fork_patches.md at the merge commit, drops the TEMPORARY marker, and gets signed off. Until then the required Attestation check cannot pass. make lint-rust runs scripts/check_fork_patches.py (Makefile:361), which refuses the marker:

$ python3 scripts/check_fork_patches.py; echo "exit=$?"
1 pin(s) not ready to land:
  - datafusion-table-providers is pinned to a branch rather than a landed revision: spiceai/datafusion-table-providers#77 has to merge first, then repoint the pin at the merge commit and drop the TEMPORARY marker
exit=1

Correctness. At this head, pushdown_round_trips --include-ignored --test-threads 1 runs 165 MongoDB, 117 DynamoDB and 56 Cosmos DB differential cases, with the two new failure-mode checks: test result: ok. 3 passed; 0 failed. Each finding in the Copilot rounds was reproduced against the database before it was fixed. For each fix, its regression check was then shown to fail with the fix reverted:

Rows compared as joined strings: fix(dynamodb, cosmosdb, mongodb): push filters down only where the source evaluates them as SQL does #14419 (comment)
A MongoDB view without _id matched the filter that matches nothing: fix(mongodb): push down only filters MongoDB evaluates as SQL does, and stop claiming a sort order datafusion-table-providers#77 (comment)
Cosmos DB single-partition reads skipped a partition key of another JSON type: fix(dynamodb, cosmosdb, mongodb): push filters down only where the source evaluates them as SQL does #14419 (comment)
DynamoDB dropped an item holding a flattened column at two paths: fix(dynamodb, cosmosdb, mongodb): push filters down only where the source evaluates them as SQL does #14419 (comment)
Limits with Inexact filters, declined with plans: fix(dynamodb, cosmosdb, mongodb): push filters down only where the source evaluates them as SQL does #14419 (comment) and fix(dynamodb, cosmosdb, mongodb): push filters down only where the source evaluates them as SQL does #14419 (comment)

#14419 (comment) is Cursor Bugbot's notice that its spend limit was reached. It asks nothing of this PR.

cursor · 2026-09-27T00:58:24Z

Bugbot is paused — on-demand spend limit reached

Bugbot uses usage-based billing for this team and has hit its on-demand spend limit.

A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue.

… from A flattened column such as m.x.y is read from the map m, a map attribute m.x, or an attribute named m.x.y, but the projection named only the first segment of the column's name. An item whose value sat under a top-level attribute with a dot in its name came back without it, so m.x.y read as NULL and a filter on it dropped the row. The projection now names every prefix of the name that ends before a dot, and the whole name. The DynamoDB round-trip suite writes an item of each kind and requires all of them to read m.x.y. The local copy shares the projection, so a round trip cannot tell.

…lan snapshot The EXPLAIN snapshot of dynamodb_timestamp_filter_pushdown still held trunk's request, which compared each timestamp bound as a string in a single offset. This branch compares the stored local date and time within a day of the bound instead, and leaves the exact comparison to the FilterExec above the scan, so the merge queue failed the test on every try. The new snapshot is the plan the test prints against amazon/dynamodb-local holding the items init_test_table writes, which is byte for byte the plan the merge queue printed against the real table.

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

lukekim · 2026-09-30T17:11:48Z

Addressed #14419 (review) as Copilot's quota-limit error on 207b2f6, which reviews nothing: "Copilot was unable to review this pull request because the user who requested the review has reached their quota limit." All four requests since 2026-09-29 20:12Z got this reply (on 1f26067, 9403797, 7c3ef2d and 207b2f6). Another request would come from the same account, so I haven't sent one. Copilot's last review with content was of 2a49ffa, with no findings: #14419 (review).

The diff hasn't changed since the approvals of 9403797. They were dismissed at 06:00Z by a rewrite of the branch history. 207b2f6 then merged trunk at 197e35d with no conflict. Against its base, the PR adds and removes the same 8803 lines at both heads. The only two lines that differ are unchanged context, next to the imports that trunk's #14412 edited:

$ diff <(git diff 940379797e^2 940379797e | grep -vE '^(index |@@)') \
       <(git diff 207b2f604a^2 207b2f604a | grep -vE '^(index |@@)')
3367d3366
<  use datafusion::prelude::SessionContext;
3369a3369
>      common::project_schema,

The merge-queue test this PR failed on should still pass after the DataFusion bump (#14430) that this merge brings in. dynamodb::dynamodb_timestamp_filter_pushdown runs only in the queue. Its snapshot has seven plan lines, and DataFusion renders six of them. Those six are byte-identical to trunk's copy of the snapshot (diff exit 0). Trunk's copy hasn't changed since #11891. It passed in merge-queue run 36736342521 on bca6fc3, which contains the bump:

PASS [   1.004s] runtime::integration dynamodb::dynamodb_timestamp_filter_pushdown

The seventh line, DynamoDBTableProviderExec { … }, is the request this branch builds from the same partial_filters. Since then, the only trunk change to the connector is #14412. It swaps SessionContext::new() for util::session_state::session_context() in the table bootstrap, and it doesn't touch that request.

Not run: the test itself on 207b2f6. It needs the AWS table whose key only CI holds, so the next queue run is that check.

Attestation stays red until remote sign-off run 36743249414 posts on 207b2f6. The run is queued, and it checks out the branch tip when it starts.

…14429 This reverts commit ce87784. The regeneration ran a binary that includes #14419, which moves several MongoDB filters out of the pushdown, so 19 of the 20 changed snapshots reflect that change rather than the null-aware NOT IN plan. They are left for their own review.

Copilot pointed out that s3[parquet]-federated and abfs_standard_versioned[parquet]-federated still carried the swapped RightAnti plan. A scan of the trunk b97f85e TPC-H sweep found nine more arms failing tpch_q16 on the same RightAnti -> null-aware LeftAnti change. Each snapshot is the new result that arm's trunk run printed. Left out: mongodb-federated (#14419 scope, as before), oracle and oracle[catalog] (older drift beyond the anti join), and mysql-arrow / postgres-arrow (their q16 plan keeps RightAnti).

Copilot AI balanced review requested due to automatic review settings September 26, 2026 16:26

lukekim requested a review from a team as a code owner September 26, 2026 16:26

github-actions Bot added area/config area/docs area/tests kind/bug Something isn't working kind/dependencies Pull requests that update a dependency file labels Sep 26, 2026

lukekim mentioned this pull request Sep 26, 2026

fix(mongodb): push down only filters MongoDB evaluates as SQL does, and stop claiming a sort order spiceai/datafusion-table-providers#77

Merged

Copilot started reviewing on behalf of lukekim September 26, 2026 16:27 View session

lukekim self-assigned this Sep 26, 2026

lukekim added the area/data-connectors label Sep 26, 2026

Copilot AI reviewed Sep 26, 2026

View reviewed changes

Comment thread crates/data-connectors/connector-cosmosdb/src/cosmosdb/provider.rs

Comment thread crates/data-connectors/connector-dynamodb/src/provider.rs

Copilot AI review requested due to automatic review settings September 26, 2026 20:48

Copilot started reviewing on behalf of lukekim September 26, 2026 20:48 View session

Copilot AI reviewed Sep 26, 2026

View reviewed changes

Comment thread crates/runtime/tests/pushdown_roundtrip.rs Outdated

Copilot AI review requested due to automatic review settings September 26, 2026 22:06

Copilot started reviewing on behalf of lukekim September 26, 2026 22:07 View session

Copilot AI reviewed Sep 26, 2026

View reviewed changes

Comment thread crates/data-connectors/connector-cosmosdb/src/cosmosdb/provider.rs Outdated

Copilot AI review requested due to automatic review settings September 26, 2026 23:13

Copilot started reviewing on behalf of lukekim September 26, 2026 23:14 View session

Copilot AI reviewed Sep 26, 2026

View reviewed changes

Comment thread crates/data-connectors/connector-dynamodb/src/filter.rs

Copilot AI review requested due to automatic review settings September 27, 2026 00:15

Copilot started reviewing on behalf of lukekim September 27, 2026 00:16 View session

Copilot AI reviewed Sep 27, 2026

View reviewed changes

Copilot AI review requested due to automatic review settings September 27, 2026 00:58

Copilot started reviewing on behalf of lukekim September 27, 2026 00:58 View session

lukekim added 2 commits September 29, 2026 22:59

Copilot AI balanced review requested due to automatic review settings September 30, 2026 06:00

lukekim dismissed stale reviews from phillipleblanc, bjchambers, and sgrebnov via 7c3ef2d September 30, 2026 06:00

lukekim force-pushed the fix/nosql-filter-pushdown branch from 9403797 to 7c3ef2d Compare September 30, 2026 06:00

Copilot started reviewing on behalf of lukekim September 30, 2026 06:00 View session

Copilot AI reviewed Sep 30, 2026

Merge branch 'trunk' into fix/nosql-filter-pushdown

207b2f6

Copilot AI balanced review requested due to automatic review settings September 30, 2026 17:00

Copilot started reviewing on behalf of lukekim September 30, 2026 17:01 View session

Copilot AI reviewed Sep 30, 2026

lukekim requested review from phillipleblanc and sgrebnov September 30, 2026 17:12

bjchambers approved these changes Sep 30, 2026

View reviewed changes

lukekim added this pull request to the merge queue Sep 30, 2026

sgrebnov mentioned this pull request Oct 1, 2026

CI: spiceai-macos-mbp-runner-01-01 fails every merge-queue job at link time (MacOSX 27.0 SDK unknown architecture), ejecting unrelated PRs #14622

Closed

This was referenced Oct 1, 2026

CI: integration test containers use fixed names, so two concurrent merge-queue batches collide on Docker 409 and eject an innocent PR #13712

Closed

expect_test.sh's first stand-in case times out on darwin-aarch64: a ~50s host stall, cause unknown #13761

Closed

Merged via the queue into trunk with commit 4a4ecc9 Oct 1, 2026
56 of 58 checks passed

lukekim deleted the fix/nosql-filter-pushdown branch October 1, 2026 09:03

phillipleblanc mentioned this pull request Oct 2, 2026

Isolate Docker integration fixtures across concurrent processes #14639

Merged

spiceemma mentioned this pull request Oct 2, 2026

docs(cosmosdb): filters, projection and limits push down into Cosmos DB spiceai/docs#2298

Merged

3 tasks done

claudespice mentioned this pull request Oct 3, 2026

test(bench): refresh the tpch_q16 explain snapshots for the null-aware NOT IN plan (fixes #13977) #14731

Merged

1 task done

krinart mentioned this pull request Oct 9, 2026

docs: document behavior that changed in v2.4.0 spiceai/docs#2332

Merged

Conversation

lukekim commented Sep 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📝 Summary

🔗 Related

📚 Docs

👀 Notes for Reviewers

Uh oh!

github-actions Bot commented Sep 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ Pull with Spice Passed

Passing checks:

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Uh oh!

cursor Bot commented Sep 26, 2026

Bugbot is paused — on-demand spend limit reached

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🔵 Needs a closer look

Uh oh!

lukekim commented Sep 27, 2026

Uh oh!

cursor Bot commented Sep 27, 2026

Bugbot is paused — on-demand spend limit reached

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

lukekim commented Sep 30, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

lukekim commented Sep 26, 2026 •

edited

Loading

github-actions Bot commented Sep 26, 2026 •

edited

Loading