v0.4.1

Release 0.4.1

v0.4.0

Release 0.4.0

• A2A SDK 0.3.12 full parity: transport enum rename, SecurityRequirement multi-key model, lenient AgentProvider decoder, artifact streaming, legacy file part support
• SDK 0.2.72 parity cleanup

v0.3.1

Release v0.3.1 — SDK 0.2.72 parity and runtime safety

v0.3.0

Full SDK 0.2.69 parity release with type safety improvements.

• InstructionsLoaded hook event with MemoryType and InstructionsLoadReason enums

• hookAgentId/hookAgentType on all HookInput variants (BaseHookInput parity)

• MessagePriority (now/next/later) on streamUserMessage

• ModelInfo gains supportsFastMode, supportsEffort, supportsAdaptiveThinking

• ToolConfig for per-tool settings (askUserQuestion preview format)

• SettingsConfig for inline settings or settings file path

• listSessions.includeWorktrees parameter

• SandboxSettings.enableWeakerNetworkIsolation

• PermissionUpdate aligned with SDK union types

• Correct SDKUserMessage shape for streamUserMessage

• Async ClaudeSession.interrupt() handling

• StringEnumJsonCodec + OpaqueStringJsonCodec utilities

• PositiveInt/PositiveDouble validated types with compile-time literal constructors

• SessionUuid validated type

• SchemaMacroSupport and JsonSchemaAst consolidation

• PermissionMode.Delegate removed (falls back to Custom("delegate"))

• RateLimitEvent fields retryAfterMs and model are now Option types

• PermissionContext.suggestions type changed to List[PermissionUpdate]

• Hook callback JS bridge signature updated for SDK 0.2.69

• @anthropic-ai/claude-agent-sdk 0.2.63 -> 0.2.69

v0.2.4

Release 0.2.4

• Upgrade to Claude Agent SDK 0.2.31
• Add debug options (debug, debugFile, strictMcpConfig) to AgentOptions
• Extend SandboxSettings with network config, ripgrep, and more fields
• Add skills field to AgentDefinition for preloading skills
• Add Setup hook event
• Add FilesPersisted system event for file persistence tracking
• Add initializationResult method to QueryStream
• Add ClaudeAIProxy MCP server type

v0.2.2

chore(release): Bump version to 0.2.2

Co-Authored-By: Claude Opus 4.5 noreply@anthropic.com

v0.2.1 - Scala 3.7.4 & Claude Code 2.1.1 Parity

feat: Add MCP wildcard permission helpers (P2) (#8)

• feat: Add MCP wildcard permission helpers

• McpToolName.wildcard(server) for server-level wildcards
• McpToolName.wildcardAll for global MCP wildcard
• McpToolNames.wildcard for instance-based wildcards

Example usage:

// Allow all tools from a server
.withAllowedTools(McpToolName.wildcard("weather-api"))

// Allow all MCP tools (use with caution)
.withAllowedTools(McpToolName.wildcardAll)

// Using McpToolNames instance
.withAllowedTools(WeatherTools.wildcard)

Co-Authored-By: Claude Opus 4.5 noreply@anthropic.com

• docs: Address PR review feedback for MCP wildcards

• Add documentation clarifying wildcard matching is handled by TypeScript SDK
• Fix comment inconsistency (mcp____ vs mcp__*)
• Add integration tests showing wildcards work with AgentOptions

Co-Authored-By: Claude Opus 4.5 noreply@anthropic.com

---

Co-authored-by: Claude Opus 4.5 noreply@anthropic.com

v0.2.0

chore: Release prep for v0.2.0

• Set default version to 0.2.0 in build.mill
• Update @anthropic-ai/claude-agent-sdk 0.1.73 → 0.1.76
• Update @types/bun 1.1.0 → 1.3.5
• Update typescript peer dep ^5 → ^5.9.3

🤖 Generated with Claude Code

Co-Authored-By: Claude Opus 4.5 noreply@anthropic.com

v0.1.0

fix: Use Task.Input for env-driven publishVersion

• Task.Input ensures PUBLISH_VERSION env var is read at runtime
• Add empty docJar to skip scaladoc (Scala.js facades cause errors)

🤖 Generated with Claude Code

Co-Authored-By: Claude Opus 4.5 noreply@anthropic.com