v0.1.0

First public release. All core language features, three runtimes at feature parity, four demo applications, and the agent adoption layer are complete.

Packages

Package	npm / pip / Go
@orcalang/orca-lang	npm install @orcalang/orca-lang
@orcalang/orca-runtime-ts	npm install @orcalang/orca-runtime-ts
@orcalang/orca-mcp-server	npm install @orcalang/orca-mcp-server
orca-runtime-python	pip install orca-runtime-python
orca-runtime-go	go get github.com/jascal/orca-lang/packages/runtime-go

Language (packages/orca-lang)

• Parser: two-phase markdown parser for .orca.md format — headings, tables, bullet lists, blockquotes. Auto-detects legacy .orca DSL files for backward compatibility
• Verifier: four-pass static analysis — structural (reachability, deadlocks, orphans), completeness (every state handles every event), determinism (mutually exclusive guards), property checking (bounded model checking with BFS: reachable, unreachable, passes_through, live, responds, invariant)
• Cross-machine verifier: cycle detection, machine resolution, child reachability, on_done/on_error event validation, combined state budget
• Compilers: XState v5 createMachine() config (TypeScript), Mermaid stateDiagram-v2
• ## effects section: declared effect types with input/output schemas; ORPHAN_EFFECT and UNDECLARED_EFFECT verifier warnings
• Machine invocation: invoke: / on_done: / on_error: bullet syntax; single-file multi-machine with --- separators
• CLI: orca verify, orca compile xstate|mermaid, orca visualize, orca actions, orca convert (legacy DSL → markdown), orca --tools --json, --stdin on all commands
• Skills (LLM-friendly structured JSON commands): /parse-machine, /verify-orca, /compile-orca, /generate-orca, /generate-orca-multi, /generate-actions, /refine-orca
• LLM integration: Anthropic, OpenAI-compatible, Ollama providers; generate_machine and refine_machine loop to convergence (up to max_iterations)
• Auth: OAuth device-code flow for Anthropic; API key via env or .orca.env
• Error catalog: 29 verifier codes documented in docs/error-catalog.md

MCP Server (packages/mcp-server)

• MCP stdio server exposing 7 tools: parse_machine, verify_machine, compile_machine, generate_machine, generate_multi_machine, generate_actions, refine_machine
• All tools accept source: string — no files required
• JSON schemas on all inputs; compatible with Claude Desktop, any MCP host

TypeScript Runtime (packages/runtime-ts)

• parseOrcaAuto — format auto-detection (.orca.md markdown or legacy DSL)
• OrcaMachine — event bus, state transitions, guard evaluation, action execution, timeout transitions, parallel regions (all-final / any-final sync), hierarchical states, child machine lifecycle
• OrcaMachine.resume() — cold-boot from snapshot without re-running on_entry
• PersistenceAdapter + FilePersistence — atomic JSONL snapshot save/load
• LogSink + FileSink / ConsoleSink / MultiSink / makeEntry() — structured JSONL audit logging
• ## effects parsing + EffectDef type

Python Runtime (packages/runtime-python)

• Feature parity with TypeScript runtime
• parse_orca_auto, OrcaMachine, decorator-style action and effect handler registration
• OrcaMachine.resume(), FilePersistence, FileSink / ConsoleSink / MultiSink

Go Runtime (packages/runtime-go)

• Feature parity with TypeScript and Python runtimes
• Goroutine-based event bus, OrcaMachine struct, guard evaluation, action registration, timeout management, parallel regions, snapshot/restore
• OrcaMachine.Resume(), FilePersistence, FileSink / ConsoleSink / MultiSink / MakeEntry()
• Module path: github.com/jascal/orca-lang/packages/runtime-go
• 16 tests

Demo Applications

• demo-ts: Playable text adventure game — 8-state machine, 4 locations, inventory, score, LLM narrative generation, MultiSink audit logging, FilePersistence snapshots
• demo-python: Agent framework — order processing (8-state workflow), multi-agent task orchestration, event bus request/response patterns
• demo-go: Ride-hailing trip coordinator — 5-machine trip.orca.md (TripCoordinator, DriverDispatch, PaymentAuth, TripExecution, FareSettlement); runs FareSettlement end-to-end with logging and persistence
• demo-nanolab: nanoGPT training orchestrator — 5-machine architecture (TrainingLab, DataPipeline, HyperSearch with parallel regions, TrainingRun, Evaluator); pluggable persistence, structured JSONL audit logging, rich terminal display, LLM workflow refinement via --refine; 47 tests (no torch required)

Documentation

• AGENTS.md — agent integration guide: installation, generation loop, LLM auth, stdin/source string patterns, multi-machine workflows, runtime extension examples (TypeScript, Python, Go)
• docs/error-catalog.md — all 29 verifier error codes with severity, cause, fix, and examples
• docs/phase-5-agent-adoption.md — Phase 6 design document
• docs/demo-ride-hailing.md — Go demo design
• docs/demo-nanolab.md — nanolab demo design
• docs/machine-invocation-design.md — machine invocation design

Test Counts

Package	Tests
orca-lang	135
runtime-ts	63
runtime-python	69
runtime-go	16
demo-nanolab	47

---