CrewAI Release Notes

What's Changed

Features

• Add XPU to OpenCLIP provider device options
• Add deepinfra as an OpenAI-compatible provider
• Record why an evaluation stopped before its verdict
• Record what the run app's Deploy met before an attempt
• Implement per-run cost and time tracking in crewai eval --models without repeated progress lines

Bug Fixes

• Fix issue where stopSequences were sent to OpenAI GPT-6/GPT-5.6/gpt-oss
• Preserve GitHub loader source attribution
• Fix SQLite flow initialization contention

Documentation

• Add snapshot and changelog for v1.15.26

Contributors

@StoyanD, @ats3v, @joaomdmoura, @kimnamu, @lorenzejay, @mhbuehler, @zouyonghe

What's Changed

Features

• Add crewai eval to print a markdown brief when run by an agent
• Queue relevant background replies
• Add experimental turn and reply identities
• Add Oracle Integrations
• Add experimental job lifecycle and runner
• Add crewai eval --models, and llm_overlay to swap models as well as roles

Bug Fixes

• Fix train filename error message in CLI
• Allow a start step to re-run on the events it listens to
• Let human_feedback emit steps expose the review in outputs
• Ensure crewai eval exits with code 1 unless the gate passed and prompt to log in when nothing was traced unattended
• Report tracing sent usage from the grant exporter
• Address multiple dependency advisories and update relevant packages

Refactoring

• Move message summarization into SummarizeMessages
• Centralize and refresh context windows

Contributors

• @Velmet44, @Vidit-Ostwal, @dependabot[bot], @fileames, @gabemilani, @github-actions[bot], @joaomdmoura, @lorenzejay

What's Changed

Features

• Add support for native Gemini 3.8 Flash
• Implement evaluation of the last traced run through AMP in crewai eval
• Record the last traced run for crewai eval instead of printing it
• Improve platform integration setup UX
• Prioritize popular platform integrations
• Enhance tracing task spans to include declared output format and results

Bug Fixes

• Fix visibility of the panel for viewing traces
• Resolve issue with turning tracing on and add an Evaluate button in the TUI
• Retry throttled provider calls in LLM
• Fall back to synchronous calls from acall in Bedrock
• Print areas graded by the evaluation in CLI
• Ensure link is shown after finalizing traces
• Keep Selenium driver reusable
• Close S3 response bodies properly
• Collapse multimodal content with shared helper
• Match roles in LLM overlay that differ only by surrounding whitespace
• Close SQLite connections in flow persistence and SQLite provider
• Preserve directory listing paths

Documentation

• Update guides for assistants to platform tools

Contributors

• @Copilot, @Mairaarshad19, @SharoonSharif, @Vidit-Ostwal, @gaoanze888, @joaomdmoura, @lorenzejay, @sclfcz, @stepchanges, @vinibrsl

What's Changed

Features

• Support aliases as connection identifiers
• Record reasons for deployment creation failures
• Collect human feedback and pause events in tracing
• Add llm_overlay context variable to route agent roles to models
• Carry task_prompt and output in agent execution payloads
• Validate platform integrations during crew setup
• Add platform tools to JSON crew wizard
• Expose CrewAI Platform application catalog
• Add OpenRouter as a supported embedding provider

Bug Fixes

• Accept CRLF in inline skill definitions
• Load text file URLs through the safe fetcher
• Close SQLite connections in kickoff task outputs storage
• Carry from_cache on the native tool path's ToolUsageFinishedEvent
• Raise ValueError instead of bare raise in uploader
• Support non-primitive types in SQLiteFlowPersistence
• Improve coding agents instructions
• Read JSON checkpoints as UTF-8
• Overwrite stale poetry.lock backup on Windows
• Key streamed tool calls by wire index in Azure
• Preserve file data content parts in Gemini
• Honor read_only on update() and recall() access times
• Send reasoning_effort to every OpenAI reasoning model
• Prevent crashes in the run TUI when streamed output contains a literal [...]
• Reject replay when stored tasks differ
• Use sys.platform guards for mypy compatibility on Windows
• Request the forced final answer as a user turn
• Keep null in the task output schema embedded in the prompt
• Align DOCXSearchTool with standard RAG fixed schema pattern

Documentation

• Add xpu to embedding device options
• List all workspace packages

Contributors

@ASTion24, @BlueX888, @HUAN2022A, @RaycarlLei, @Rohitkanithi, @SWAPI03, @SharoonSharif, @ShivangiRay, @Theater-ahyeon, @Vidit-Ostwal, @gamal1osama, @github-actions[bot], @joaomdmoura, @lorenzejay, @mhbuehler, @modusensus, @monkscode, @Rohitkanithi, @roli-lpci, @vinibrsl, @wangtaotaotao95

What's Changed

Features

Add telemetry to track checkpoint runtime and CLI usage.

Bug Fixes

Fix gateway errors reported inside an HTTP 200 response.

Keep deploy push on the AMP create source.

Report scrape failures instead of raising IndexError in the Oxylabs integration.

Route all DashScope models through the native provider.

Persist JSON checkpoints as UTF-8.

Drop stray $ in f-string search URLs for BrightData.

Support list-form type arrays in JSON schema conversion.

Correct gpt-4o-mini context window from 200000 to 128000.

Preserve streaming tool call arguments at contentBlockStop.

Make pre-commit hooks portable on Windows.

Documentation

Clarify that tracing is managed separately from telemetry.

Fix parallel search reference link.

Remove stale parameters from handle_llm_stream_chunk docstring.

Fix streaming output docstring examples.

Use organization UUIDs in the skill install reference.

Contributors

@Alphaxiaoteng, @DrewWhittleNZ, @Ghraven, @Shxiao101, @Vaishnavi220506, @Vidit-Ostwal, @danielfsbarreto, @georgeatparallel, @github-actions[bot], @jessemiller, @joaomdmoura, @kimnamu, @kiwoongyoon, @liang0417, @lorenzejay, @oxy-giedrius, @simpleqt, @uoparaji

What's Changed

Features

• Add Clipper integrations client
• Add now() to the CEL expression environment
• Record how a crew run ended for every user
• Report machine size as a coarse band, not a core count
• Add injectable client for CrewAI platform tools

Bug Fixes

• Fix reading of octet-stream and xlsx URLs in urlreadtool
• Fix appending trailing user turn in native Gemini provider
• Normalize scheme and port in Ollama base URL
• Preserve reusable scope configs in memory
• Bump pypdf to 6.16.2 for security vulnerability
• Bump nltk to 3.10.3 for security vulnerability
• Fix native structured outputs for current Claude models and Snowflake CVE floor
• Run model call hooks on every path and propagate a deny

Documentation

• Remove CodeInterpreterTool from AI/ML overview examples
• Point prompt-template link at its current path
• Update channels guide to current CopilotKit channels API
• Refresh retired Gemini model IDs

Contributors

@Vidit-Ostwal, @a-yeyang, @github-actions[bot], @hvlcrs, @joaomdmoura, @kikifrost, @lorenzejay, @lucasgomide, @parthiban-sivakumar, @ranst91, @tandede, @thiagomoretto, @vinibrsl

What's Changed

Features

• Promote conversational flows to stable
• Record a created deployment with the given UUID
• Enhance conversational flow documentation and APIs
• Let a declaration name the router's response format
• Let a chat flow declare its own state shape
• Accept crew-style LLM config in a conversational declaration
• Report project creation with the minted ID
• Record whether a run had inputs, without recording the inputs
• Backfill project ID from every user-invoked project command

Bug Fixes

• Preserve tool results when the final answer is empty
• Map default Claude Sonnet 4.6 to its 1M context window
• Raise Anthropic default max_tokens for large tool calls
• Render message content parts as text, not as a Python repr
• Keep message roles when Agent.kickoff gets a conversation
• Skip interception hooks on crewai-internal flows
• Record task failures as failures, not successes
• Emit the flow lifecycle on a suppressed resume
• Open the conversational TUI for a declarative chat flow
• Record crew_memory as a string, not a bool
• Always emit project_id so absent and empty stay distinct

Documentation

• Clarify Arize Phoenix observability docs

Contributors

• @Vidit-Ostwal, @arizedatngo, @joaomdmoura, @lorenzejay, @lucasgomide

What's Changed

Features

• Add declarative conversational flows documentation
• Synthesize built-in conversational methods for declarations
• Enable declarations to drive conversational mode
• Make conversational opt-in unmistakable
• Carry the AMP slug on tools resolved from a slug reference
• Handle oversized single messages during chunking

Bug Fixes

• Fix usage of the URL hostname as MCP HTTP and SSE server_name
• Close the agent scope on every failed attempt
• Attribute tool errors to the tool that failed
• Pin SSRF checks to each redirect hop and peer IP
• Resolve issues with native tool calls broken over OpenAI Responses API

Documentation

• Update documentation with a snapshot and changelog for v1.15.16

Contributors

• @Copilot, @Vidit-Ostwal, @github-code-quality[bot], @joaomdmoura, @lorenzejay, @lucasgomide, @theCyberTech

What's Changed

Features

• Introduce execution context management with UUID support
• Record what kind of exception ended a flow
• Record when a trace batch is shared with AMP
• Count deployments from any origin and record where they started

Bug Fixes

• Record the running release on every emitted span
• Fix MySQL search table name validation
• Stop a failed turn from marking the next one as failed

Documentation

• Add Frontend guides for CopilotKit and AG-UI

Contributors

• @joaomdmoura, @lorenzejay, @ranst91, @theCyberTech

What's Changed

Features

Report flow outcome, duration, and human-in-the-loop signals.

Bug Fixes

• Emit FlowStartedEvent when a boundary hook aborts the flow.
• Scope span export to our own tracer provider.
• Bump torch to version 2.13.0 to address security vulnerability.
• Bump gitpython to version 3.1.58 in crewai-tools[github].

Refactoring

• Update date injection functionality in agents.
• Standardize CLI flags to kebab-case.

Documentation

Snapshot and changelog for v1.15.14.

Contributors

@Vidit-Ostwal, @joaomdmoura, @lorenzejay, @lucasgomide, @theCyberTech

What's Changed

Bug Fixes

Fix preservation of provider on LiteLLM-routed models.

Harden brittle LLM event-bus mocks.

Fix underreporting of Anthropic cache token usage.

Bump h2 to version 4.4.1 to address security vulnerability GHSA-6hr6-w5qg-qmwg.

Documentation

Add DOCS_TRANSLATIONS workflow for locale synchronization.

Fix broken README links, table of contents, and contribution guidance.

Snapshot and changelog for version 1.15.12.

Contributors

@Vidit-Ostwal, @joaomdmoura, @lucasgomide, @theCyberTech