What CrewAI shipped
Each line links to CrewAI's own note; the short summaries are FoxPlug's. Not affiliated with CrewAI.
Last 90 days, as of November 21, 2026
Changes to act on (last 90 days)
- November 11, 2026 · Removal · Removal: Release 1.15.40 was released on 11 November
CrewAI ships more often than 62% of the other 8 ai infrastructure products we track: 4th of 9, at 2.2 entries a week over the days its feed covers. How the rank is worked out.
November 2026
November 2026 brought artifact versioning, optional ChromaDB with local Qdrant defaults, spreadsheet reading, experimental conversational sessions, and fixes across flows and caching.
- November 19, 2026 · 1.15.44
Adds saving files as versioned, shareable artifacts. Fixes include action reference resolution, chunk separator duplication, nested defaults, image format restrictions, cache eviction order, GitHub docs paths, train exit status, av pinning, and FilePath contai
- November 19, 2026 · 1.15.43
- November 18, 2026 · 1.15.42
ChromaDB becomes an optional extra, with knowledge storage defaulting to local Qdrant and decoupled from ChromaDB. Adds tool selection for standalone agents plus numerous fixes for output serialization, flow IDs and timeouts.
- November 12, 2026 · 1.15.41
Private URL hosts can now be approved. Projects depend on PyMuPDF 2.x, earlier step outputs persist across a human-feedback resume, and the
cache_breakpointmarker is stripped on the LiteLLM path. - November 11, 2026 · 1.15.40
ChromaDB is removed from the embedding function layer, historical state can be looked up by row id, JSONL joins RAG data type discovery, and standalone agents scaffold as JSON by default. Fixes cover async
after_llm_callhooks and stopping agents with unresolv - November 9, 2026 · 1.15.39
Documentation now covers the
apps_not_connectederror on the /kickoff and /status endpoints. - November 8, 2026 · 1.15.38
- November 7, 2026 · 1.15.37
Agents can read spreadsheets through crewai-files, and standalone agent projects can be scaffolded and loaded.
- November 6, 2026 · 1.15.36
Introduces experimental conversational agent sessions. crewai run now asks for a missing API key and new projects get the chosen provider's extra.
- November 3, 2026 · 1.15.35
- November 1, 2026 · 1.15.34
Adds crewai skill search, passing --inputs to Python flows, and deploy create --zip with commit reporting. Bug fixes cover missing API key errors, typed errors after an empty LLM answer, JSON output refactoring, and dependency floors.
October 2026
CrewAI shipped CLI and eval additions, new providers and integrations, skill search and flow inputs, plus fixes across tracing, JSON loading and tests.
- October 29, 2026 · 1.15.33
Adds skill search, flow inputs via crewai run, and zip deploys. Fixes cover LiteLLM token usage reporting, JSON crew loading, tracing env handling, and crewai update printing upgrade commands instead of rewriting projects.
- October 27, 2026 · 1.15.31
- October 27, 2026 · 1.15.32
Fixes flaky tests and updates the snapshot and changelog for v1.15.31.
- October 24, 2026 · 1.15.30
Fixes AGENTS.md so the login process starts during evaluation and stops after three rounds. Documentation snapshot and changelog updated for v1.15.29.
- October 23, 2026 · 1.15.29
Adds LiteLLM-compatible kwargs, CLI login organization selection, Azure DocumentDB vector search, and promotes AgentExecutor out of experimental status. Eval gains export and single-command verdict options.
- October 14, 2026 · 1.15.28
- October 9, 2026 · 1.15.27
Adds deepinfra as an OpenAI-compatible provider, XPU device options for OpenCLIP, and per-run cost and time tracking in eval. Fixes stopSequences sent to newer OpenAI models and SQLite flow contention.
- October 8, 2026 · 1.15.26
Fixes rwlock ownership retention when acquisition is interrupted, documentation site source URL preservation, and task output display in log-tasks-outputs.
- October 7, 2026 · 1.15.24
Adds crewai eval outputting a markdown brief, a --models flag with
llm_overlay, Oracle integrations and experimental job lifecycle support. Includes CLI and tracing fixes. - October 7, 2026 · 1.15.25
Documentation-only release adding the snapshot and changelog for v1.15.24.
September 2026
- September 28, 2026 · 1.15.23
Adds native Gemini 3.8 Flash support and lets crewai eval score and record the last traced run through AMP. Includes retry for throttled provider calls and TUI tracing fixes.
- September 16, 2026 · 1.15.22
Added aliases as connection identifiers, deployment failure reasons, an
llm_overlaycontext variable, platform tools in the JSON crew wizard, and OpenRouter embeddings. - September 9, 2026 · 1.15.21
Added telemetry for checkpoint runtime and CLI usage, with fixes for gateway errors returned in HTTP 200 responses, JSON checkpoint encoding and Windows pre-commit hooks.
- September 4, 2026 · 1.15.19
Adds Clipper integrations, a now() function for CEL expressions and an injectable client for platform tools. Fixes include URL reading, Ollama base URL normalization and security bumps to pypdf and nltk.
- September 4, 2026 · 1.15.20
Fixes discovery of legacy platform tool aliases.
August 2026
- August 27, 2026 · 1.15.18
Promotes conversational flows to stable with enhanced documentation and APIs, and lets chat flows declare their own state shape. Fixes include preserving empty-answer tool results and Anthropic token limits.
- August 20, 2026 · 1.15.17
Adds documentation for declarative conversational flows, which declarations can now drive. Fixes include per-hop SSRF checks and native tool calls over the OpenAI Responses API.
- August 14, 2026 · 1.15.16
Introduces execution context management with UUID support and records the exception type that ended a flow. Adds frontend guides for CopilotKit and AG-UI.
- August 12, 2026 · 1.15.15
- August 8, 2026 · 1.15.14
Splits runtime context from the coding agent and adds a project ID.
- August 7, 2026 · 1.15.13
Fixes provider preservation on LiteLLM-routed models, Anthropic cache token underreporting and brittle event-bus mocks. Bumps h2 to address a security vulnerability.
- August 5, 2026 · 1.15.12
No release notes were provided for this version.
Ship updates like these to your users
CrewAI keeps a public record of what it ships. FoxPlug gives your product the same: a changelog page, an in-app widget and release posts, all from one place.
See FoxPlugSee a live changelog and widgetYour changelog, written for you
FoxPlug writes your changelog from your releases and commits in plain English and publishes it as a page and a feed your users can follow. The page above shows what that looks like for a product people really use.
Write my changelog See the changelog FoxPlug wrote from its own commits
Free to start. Pro is a month.
FoxPlug Pro
- A public changelog page and feed for each project
- Written from your releases and commits, in plain English your users read
- Connect GitHub and new releases become new entries