v0.2.0

What's Changed

• docs: control flow runs end-to-end on the engine (#259 shipped) by @leo-aa88 in #279
• ci(release): dispatch/tag-only release, no auto-cut on merge to main by @leo-aa88 in #280
• test(engine): fix flaky TestRun_approvalStep_parallelSuspendsOnlyItsBranch (#276) by @leo-aa88 in #281
• fix(execir): resume concurrent multi-gate HITL across decisions, not fail-closed (#275) by @leo-aa88 in #282
• fix(plan): fold execir program digest into run-start/resume drift hash (#277) by @leo-aa88 in #283
• feat(engine): retire the WorkflowStep DAG runtime — execir is the single run path (#278) by @leo-aa88 in #284
• docs: #285 plan + ADR 002 §6 (bounded iteration) by @leo-aa88 in #295
• docs: refresh README + examples index for shipped state by @leo-aa88 in #296
• feat(lang,execir): bounded while cond limit N (#288, #289) by @leo-aa88 in #299
• feat(lang): multiline strings + instructions agent field (#286, #287) by @leo-aa88 in #298
• feat(engine,execir): durable bounded while — resume without duplicate effects (#290) by @leo-aa88 in #300
• feat(spec,engine): multiple autonomous operations per Tool grant (#291) by @leo-aa88 in #301
• feat(check): lower agent input/output type refs into AgentSpec schema fields (#294) by @leo-aa88 in #302
• feat(lang): allow passing the whole workflow input to an agent (#303) by @leo-aa88 in #304
• feat(spec,lang): relax executable-graph validation for synthetic control-flow steps (#305) by @leo-aa88 in #306
• feat(examples): flagship implement-review-loop — bounded control around nondeterministic agents (#292) by @leo-aa88 in #307
• feat(models): add Grok, Gemini, and Kimi model adapters by @leo-aa88 in #308
• feat(plan,execir): surface per-agent invocation bounds from bounded loops (#293) by @leo-aa88 in #309
• feat(lang): add description and constraints fields to the .agent surface (#310) by @leo-aa88 in #311
• docs(examples): migrate examples to lead on .agent (#297) — batch 1: audit-tamper by @leo-aa88 in #312
• docs(examples): migrate examples to .agent (#297) — batch 2: hitl-resume, policy-denial-midrun, regression-test by @leo-aa88 in #313
• docs(examples): migrate examples to .agent (#297) — batch 3 (final): env-overlays, multi-agent, incident-triage by @leo-aa88 in #314
• docs(readme): lead with a plain-English hook, quickstart, and a one-minute example by @leo-aa88 in #315
• feat(lang): ${...} string interpolation in .agent argument values (#316) by @leo-aa88 in #317
• docs: correct #204 closed-world status — dispatch-time enforcement is shipped by @leo-aa88 in #318
• examples: migrate the pr-review family to .agent by @leo-aa88 in #319
• docs: run-pinned manifest at resume is shipped (#207), not pending by @leo-aa88 in #320
• chore: rename module/org LAA-Software-Engineering → Terfyn by @leo-aa88 in #322
• build(deps): bump go.opentelemetry.io/otel/exporters/stdout/stdouttrace from 1.45.0 to 1.46.0 by @dependabot[bot] in #321
• feat(tools): native workspace adapter (read_file / write_file / run_tests) by @leo-aa88 in #324

Full Changelog: v0.1.105...v0.2.0