v0.6.66: child trace spans, reranker controls, attachment previews, logs block, parallel-in-loop wall clock, gpt-image-2

[waleedlatif1]

• fix(table): return 400 instead of 500 for malformed sort/filter input (fix(table): return 400 instead of 500 for malformed sort/filter input #4425)
• fix(knowledge): revert column width multipliers that misaligned Name header (fix(knowledge): revert column width multipliers that misaligned Name header #4427)
• fix(executor): strip childTraceSpans from block state before LLM tool calls (fix(executor): strip childTraceSpans from block state before LLM tool calls #4428)
• feat(knowledge): expose Cohere reranker controls (feat(knowledge): expose Cohere reranker controls #4429)
• fix(mothership): catch draft restore errors instead of crashing /home (fix(mothership): catch draft restore errors instead of crashing /home #4433)
• fix(mothership): stop persisting log resources from get_workflow_logs and self-heal stale log panel entries (fix(mothership): stop persisting log resources from get_workflow_logs and self-heal stale log panel entries #4424)
• feat(mothership): restore attachment previews on draft and add video support (feat(mothership): restore attachment previews on draft and add video support #4435)
• fix(copilot): redact sim_key API keys from persisted Mothership chat messages (fix(copilot): redact sim_key API keys from persisted Mothership chat messages #4434)
• feat(image-generator): add gpt-image-2 model support (feat(image-generator): add gpt-image-2 model support #4437)
• fix(logs): split summary/detail contracts to make trace tab gate type-safe (fix(logs): split summary/detail contracts to make trace tab gate type-safe #4431)
• fix: double wrap response of guest session handler (fix: double wrap reponse of guest session handler #4438)
• improvement(mothership): streaming state transitions (improvement(mothership): streaming state transitions #4439)
• feat(logs): add Logs block for querying execution logs from workflows (feat(block): add Logs block for querying execution logs from workflows #4442)
• fix(terminal): use wall-clock duration for loop iterations with concurrent children (fix(terminal): use wall-clock duration for loop iterations with concurrent children #4443)
• fix(tables): suppress phantom rows on sort, center gutter numbers, stop select-all viewport jump (fix(tables): suppress phantom rows on sort, center gutter numbers, stop select-all viewport jump #4445)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		May 5, 2026 3:56am

[cursor]

PR Summary

Medium Risk
Medium risk: changes touch auth/session responses, log listing/detail APIs (new pagination + sorting), and vector search reranking inputs, which can affect data access patterns and UI behavior. Most changes are additive/contract-driven but could introduce subtle regressions in pagination ordering, auth envelopes, or search cost/behavior.

Overview
Adds new capabilities across docs, APIs, and Mothership UX, including GPT Image 2 support, Table trigger docs, and expanded tool documentation for Knowledge/Mem0.

On the backend, introduces a cursor-based, multi-sort logs listing in GET /api/logs, refactors GET /api/logs/[id] onto shared contracts + fetchLogDetail, and adds GET /api/logs/by-execution/[executionId]; also updates logs auth to support checkSessionOrInternalAuth. Knowledge search now exposes reranker input sizing (rerankerInputCount) and optional BYOK reranker API key passthrough, with guardrails/cost-bounding.

In Mothership/Copilot, improves streaming state propagation by including streamId in task status events and stop notifications, returns chatId in stream batch resume responses, hardens draft restore (prevents crashes and restores attachment previews), adds video attachment previews, and supports redacted credential display via SecretReveal. Table row APIs now return 400 for malformed sort/filter SQL validation errors instead of generic 500s.

^{Reviewed by Cursor Bugbot for commit 51addc5. Configure here.}

[greptile-apps]

Greptile Summary

• Introduces a Logs block and three log tools (logs_query, logs_get, logs_get_execution), a sim_key redaction layer for Mothership chat (write-side server redaction + read-side page-session key cache), and GPT Image 2 support with per-model size/quality/background validation.
• Fixes wall-clock duration for loop iterations containing concurrent parallel branches by switching from sum(durationMs) to max(endedAt) − min(startedAt), backed by comprehensive new tests. Also strips childTraceSpans from the block output passed to LLM tool-call state.
• Addresses several correctness bugs: double-wrapped anonymous session response, 500→400 for malformed table filter/sort input, draft restore crash, and cross-table cursor logic in the log list route.

Confidence Score: 5/5

Safe to merge — no P0 or P1 issues found across 81 changed files.

All fixes are well-scoped and tested: wall-clock duration has regression tests, redaction has unit tests, anonymous session fix is confirmed by updated route tests, and error-status mapping follows established patterns. No identified correctness or security defects.

No files require special attention.

Important Files Changed

Filename	Overview
apps/sim/lib/copilot/chat/sim-key-redaction.ts	New module: two-sided sim_key redaction — server-side strip before DB persist, client-side page-session cache for live key re-injection. Well-tested and handles multi-block streaming merge.
apps/sim/executor/execution/block-executor.ts	Strips childTraceSpans from block output before setBlockOutput and return value; error path now reads spans directly from ChildWorkflowError. Both paths are consistent.
apps/sim/app/workspace/[workspaceId]/w/[workflowId]/components/terminal/utils.ts	Wall-clock fix: subflow and iteration durations now use max(endedAt)−min(startedAt) instead of sum(durationMs), preventing inflation for concurrent parallel branches. Tests added.
apps/sim/blocks/blocks/image_generator.ts	Adds gpt-image-2 model with per-model size/quality/background ALLOWED_ maps for safe param validation; follows existing duplicate-id conditional subblock pattern for shared quality/size fields.
apps/sim/lib/table/sql.ts	Introduces TableQueryValidationError class; all validation throws upgraded so routes can map caller errors to 400 while preserving 500 for programmer errors.
apps/sim/lib/logs/fetch-log-detail.ts	New shared loader for the log-detail shape used by both by-id and by-execution routes; correctly enforces workspace scoping and permission JOIN.
apps/sim/blocks/blocks/logs.ts	New Logs block wiring query/get_log/get_execution operations to their respective tools; param mapping and validation look complete.
apps/sim/tools/logs/query.ts	New logs_query tool; correctly reads workspaceId from execution context and builds query string. Handles empty/falsy limit and null nextCursor.
apps/sim/lib/copilot/chat/persisted-message.ts	Wires sim-key redaction into message persistence: redactSensitiveContent on content, redactToolCallResult on tool results, mergeAndRedactPersistedBlocks on content blocks.

Sequence Diagram

sequenceDiagram
    participant C as Client (useChat)
    participant S as /api/copilot/chat/stream
    participant P as buildPersistedAssistantMessage
    participant DB as Postgres
    participant R as restoreRevealedSimKeysForMessage

    C->>S: SSE stream
    S-->>C: token chunks (incl. credential tags)
    Note over C: captureRevealedSimKeys()<br/>stores live key in revealedSimKeysRef
    S->>P: assemble persisted message
    P->>P: redactSensitiveContent()<br/>mergeAndRedactPersistedBlocks()
    P->>DB: persist (key → REDACTED)
    C->>DB: refetch after stream ends
    DB-->>C: message with redacted tags
    C->>R: restoreRevealedSimKeysForMessage(msg, cache)
    R-->>C: message with live key re-injected (in-memory only)

Loading

_{Reviews (2): Last reviewed commit: "fix(terminal): use wall-clock duration f..." | Re-trigger Greptile}

[waleedlatif1]

@greptile

[waleedlatif1]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 51addc5. Configure here.}