Vidbyte SDK v0.1.0

Alpha release: Vidbyte SDK APIs may change between minor versions.

This is the first public PyPI release of Vidbyte SDK. It includes the clean-install packaging fix, modern package metadata, trusted publishing, repository cleanup, and public contribution/security guidance.

What's Changed

• fix: resolve review comments from Vidbyte PR #187 by @cerredz in #1
• feat: Custom Function Tools by @cerredz in #7
• fix: resolve SDK review comments from PR #17 by @cerredz in #18
• feat: simplify prompt interface by @cerredz in #19
• feat: Agent Tool API Consolidation by @cerredz in #23
• feat: expand prompt descriptions to 6-8 sentences for all 15 strategies by @cerredz in #22
• fix: resolve SDK review comments from PR #21 by @cerredz in #25
• fix: expand prompt strings to 6-8 sentences for all 15 strategies by @cerredz in #26
• feat(pipelines): add SequentialPipeline, ParallelPipeline, ConditionalPipeline by @cerredz in #24
• docs: add Vidbyte SDK doc skill by @cerredz in #28
• fix: resolve SDK review comments from PR #29 by @cerredz in #31
• feat: add agent runtime middleware by @cerredz in #32
• fix: resolve review comments from PR #33 by @cerredz in #34
• feat: add MapReducePipeline by @cerredz in #36
• feat: add prompt-engineering master prompt family by @cerredz in #41
• fix: resolve review comments from PR #35 by @cerredz in #38
• feat: Agent Strategy Chain Execution by @cerredz in #40
• fix: resolve SDK review comments from PR #39 by @cerredz in #43
• fix: resolve review comments from PR #42 by @cerredz in #44
• feat(tracing): agent tracing and observability layer by @cerredz in #37
• fix: resolve SDK review comments from PR #52 by @cerredz in #53
• feat: MCP Studio Server - expose SDK capabilities as MCP tools by @cerredz in #47
• docs: resolve context window algorithm guide comments from PR #53 by @cerredz in #54
• feat: add five new context compaction modes by @cerredz in #51
• refactor(mcp): McpSchema class, server/ package with per-handler files, and mcp-server skills by @cerredz in #55
• docs: add guide for installing and registering the Vidbyte SDK as an MCP server by @cerredz in #56
• feat: MCP discovery tools — SearchMcpServersTool and AttachMcpServerTool by @cerredz in #57
• fix: resolve review comments from PR #58 by @cerredz in #68
• fix: resolve review comments from PR #68 by @cerredz in #69
• feat: context window primitives by @cerredz in #61
• feat: Resolve PR 60 Comments by @cerredz in #70
• feat(middleware): add five new prebuilt middleware builtins by @cerredz in #64
• feat: Preset MCP Servers Catalog & One-Line Attachment by @cerredz in #65
• fix: resolve review comments from PR #62 by @cerredz in #71
• fix: resolve review comments from PR #65 by @cerredz in #73
• feat: Security Middleware — Canary Tripwire, Confused Deputy Guard, Honeypot Tool by @cerredz in #67
• feat: First-Class SDK Evals Harness (vidbyte.evals) by @cerredz in #59
• feat(mcp-presets): expand preset catalog from 101 to 201 (+100 new presets) by @cerredz in #76
• feat: Memory Provider Tools (Supermemory, Mem0, Zep, Cognee, Letta) by @cerredz in #77
• refactor: remove strategies abstraction from vidbyte-sdk by @cerredz in #78
• feat: Resolve PR #77 Review Comments by @cerredz in #80
• feat: Context Window Templates — deterministic slot tracing for CW algorithms by @cerredz in #81
• feat: add AudioModelRunner, EmbeddingModelRunner, and StreamingTextModelRunner by @cerredz in #82
• feat: Smart Truncation of Tool Results Compaction by @cerredz in #83
• feat: Custom Exception Constructors by @cerredz in #86
• fix(filesystem): migrate filesystem tools to BaseTool by @cerredz in #87
• fix(middleware): isolate per-run state to prevent concurrent agent run corruption by @cerredz in #88
• fix: resolve review comments from PR #84 by @cerredz in #92
• feat(transport): async HTTP transport via httpx — fix ParallelPipeline concurrency by @cerredz in #89
• docs: Update README Agent Modality Guidance by @cerredz in #97
• refactor(agents): consolidate runner callbacks into RunnerHandle by @cerredz in #99
• feat(structured-output): add output_schema to agents and tools with provider-native enforcement by @cerredz in #101
• fix: resolve SDK review comments from PR #102 by @cerredz in #104
• feat(tools): TrajectoryCheckpointTool and ReflexionTool — context-window algorithms as model-callable tools by @cerredz in #112
• fix: resolve review comments from PR #110 by @cerredz in #115
• fix: resolve SDK review comments from PR #105 by @cerredz in #116
• fix: resolve SDK review comments from PR #106 by @cerredz in #117
• docs(skills): audit & update all skills after recent merges by @cerredz in #121
• fix: Harden LangSmith tracing live delivery by @cerredz in #123
• docs: expand README SDK feature coverage by @cerredz in #128
• feat: Deterministic Compaction Middleware Expansion by @cerredz in #124
• fix: resolve review comments from PR #120 by @cerredz in #129
• fix: Preserve chat system prompts and trace input messages by @cerredz in #130
• fix: stop agent loop on plain assistant output by @cerredz in #132
• fix(tracing): LangSmith trace closure bugfixes (flush, trace_id threading, metadata filter) by @cerredz in #133
• feat: trace-backed compaction methods by @cerredz in #126
• fix: resolve review comments from PR #127 by @cerredz in #134
• docs: expand GitHub content READMEs by @cerredz in #138
• docs: add llms documentation bundle by @cerredz in #139
• fix: resolve review comments from PR #148 (AgentLoopSettings) by @cerredz in #149
• feat: Aggregate Agent (Mixture-of-Agents) by @cerredz in #140
• fix(tracing): close root trace and llm.call span on BaseException cancellation (#142) by @cerredz in #150
• docs(prompts): add Available Prompts catalog to prompts README by @cerredz in #154
• fix(runtime): insert assistant tool-call turn before tool results in conversation history (#144) by @cerredz in #152
• feat(middleware): repeated-output loop detection with soft/hard thresholds (resolves review on #151) by @cerredz in #153
• feat(prompts): add templates.master prompt-authoring system prompt by @cerredz in #155
• feat(prompts): add 4 sections to templates.master anatomy by @cerredz in #156
• feat(agents): add run_sequentially and arun_sequentially to BaseAgent by @cerredz in #157
• feat: Agent Behavior Facade (agent.behavior) by @cerredz in #160
• feat: add agent output behavior predicates by @cerredz in #162
• fix: resolve review comment from PR #157 by @cerredz in #159
• feat: Eval Template Bundles by @cerredz in htt...