v0.11.4: OpenAI Hotfix

This release includes a hotfix for a bug in the OpenAI client: #390

Full Changelog: v0.11.3...v0.11.4

v0.11.3: Exception-al Service

What's Changed

New Features 🎉

• Update pretty-print setting name by @jlowin in #375

Fixes 🐞

• Remove model typing due to validation error by @jlowin in #376
• fix(orchestrator): OrchestratorError to serialize Exception as str by @teocns in #373

Full Changelog: v0.11.2...v0.11.3

v0.11.2: Great Success

This is a small release that tweaks mark_success instructions to improve GPT-4o mini responses.

What's Changed

Other Changes 🦾

• Improve instructions generation for success tool by @jlowin in #372

Full Changelog: v0.11.1...v0.11.2

v0.11.1: Let That Async In

What's Changed

New Features 🎉

• Add async handlers by @jlowin in #364

Docs 📚

• add syntax highlighting by @jlowin in #357
• Fix broken link and broken code in docs by @discdiver in #359

Other Changes 🦾

• Update test_history.py by @AranavMahalpure in #360
• add slackbot example and update typing + debug logs by @zzstoatzz in #363
• fully rm invalid test by @zzstoatzz in #366
• Add AI labeler by @jlowin in #367
• Pass basemodel attributes directly as kwargs by @jlowin in #369
• Orchestrator: exclude handlers field from being serialized by @teocns in #370

New Contributors

• @AranavMahalpure made their first contribution in #360
• @teocns made their first contribution in #370

Full Changelog: v0.11.0...v0.11.1

v0.11.0: Flow Stopper

The 0.11 release includes new ways of controlling the agentic loop to introduce custom early termination conditions.

What's Changed

New Features 🎉

• Introduce ways to control agentic loop termination by @jlowin in #348

Enhancements 🚀

• Improve type hints by @jlowin in #343
• Improve task rendering by @jlowin in #344
• Allow parent tasks to optionally wait for subtasks by @jlowin in #346
• Allow model_kwargs to be passed to llm API by @jlowin in #347
• Add LLM rules for custom or unrecognized models by @jlowin in #350
• Add compilation flag for removing all system messages by @jlowin in #355

Fixes 🐞

• Show all data in live even when taller than terminal by @jlowin in #345

Breaking Changes 🛫

• Make flow context explicit by @jlowin in #354

Docs 📚

• "Added in" → "New in" for docs badge by @jlowin in #339
• chore: update settings.py by @eltociear in #352

Other Changes 🦾

• Remove Prefect overrides by @jlowin in #311

Full Changelog: v0.10.0...v0.11.0

v0.10.0: Total Recall

The 0.10 release is headlined by a new memory system, so that agents can retain partitioned knowledge across flows and invocations. Agents can have multiple memory modules, each with its own instructions and shared access patterns. Memory is backed by a pluggable provider interface; 0.10 includes support for Chroma and LanceDB.

What's Changed

New Features 🎉

• remove old typer extra and update json -> model_dump_json (LC 0.3+ support) by @zzstoatzz in #312
• Support langchain 0.3 by @jlowin in #321
• Add llm-specific prompt instructions by @jlowin in #322
• Add agent memories by @jlowin in #326
• Bump langchain versions by @jlowin in #328
• Allow completion tools to be customized per-task by @jlowin in #330
• Add chroma cloud configs by @jlowin in #331
• Add LanceDB memory provider by @jlowin in #334
• Support async tasks with task decorator by @jlowin in #337

Fixes 🐞

• Ensure flow decorator works with async flows by @jlowin in #336

Breaking Changes 🛫

• Change raise_on_error to raise_on_failure by @jlowin in #327

Docs 📚

• Add docs for multiple labels by @jlowin in #320
• Remove comma from pip install by @ahuang11 in #324
• Add version badges by @jlowin in #329
• Add memory example by @jlowin in #332
• Update all examples by @jlowin in #335

Other Changes 🦾

• Fix spacing by @jlowin in #333

New Contributors

• @ahuang11 made their first contribution in #324

Full Changelog: v0.9.4...v0.10.0

v0.9.4: Pin It to Win It

What's Changed

Other Changes 🦾

• Pin below langchain core 0.3 by @jlowin in #318

Full Changelog: v0.9.3...v0.9.4

v0.9.3: Model Behavior

What's Changed

New Features 🎉

• Support automatic model configuration on agents by @jlowin in #308
• Improve gemini support by @jlowin in #309

Full Changelog: v0.9.2...v0.9.3

v0.9.2: Handle with Care

This release exposes and documents Handlers for users to customize. Handlers provide a way to observe and react to events that occur during task execution. They allow users to customize logging, monitoring, or take specific actions based on the orchestration process.

What's Changed

New Features 🎉

• Expose and document handlers by @jlowin in #307

Docs 📚

• Add example doc: multi-llm workflows by @jlowin in #306

Full Changelog: v0.9.1...v0.9.2

v0.9.1

This small release ensures that users have Prefect's ephemeral mode enabled by default.

What's Changed

Other Changes 🦾

• Use prefect ephemeral mode by @jlowin in #305

Full Changelog: v0.9...v0.9.1