Spark v0.2.0

What's Changed

• 🔧 chore: align cue and graft tool contracts by @zrr1999 in #2
• ✨ feat: merge Navia cockpit into Spark by @zrr1999 in #3
• ✨ feat: harden durable session workbench flows by @zrr1999 in #5
• ✨ feat: propagate explicit model and add synchronous wait by @zrr1999 in #6
• ✨ feat: native side threads and architecture guardrails by @zrr1999 in #7
• 🐛 fix: keep pnpm hooks read-only by @zrr1999 in #8
• ♻️ refactor(architecture): converge Spark package ownership by @zrr1999 in #10
• ♻️ refactor(tooling): converge package script grammar by @zrr1999 in #11
• ✨ feat(daemon): own autonomous driver runtime by @zrr1999 in #9
• ✅ test: enforce behavior-first contracts by @zrr1999 in #12
• ✨ feat(runtime): require structured role outcomes by @zrr1999 in #13
• 🐛 fix(runtime): dedupe memory checkpoint delivery by @zrr1999 in #14
• 🐛 fix(session): make daemon transcript authoritative by @zrr1999 in #16
• 🐛 fix(runtime): harden daemon and Cockpit stability by @zrr1999 in #17
• 🐛 fix(cockpit): count only product artifacts by @zrr1999 in #19
• ♻️ refactor(cockpit): remove duplicate message tab by @zrr1999 in #20
• 🐛 fix(cockpit): stabilize preview dependency loading by @zrr1999 in #21
• ✨ feat(update): add managed release updater by @zrr1999 in #18
• 📝 docs: add bilingual Spark user documentation site by @zrr1999 in #22
• 🐛 fix(daemon): rebuild source service artifact by @zrr1999 in #23
• 🐛 fix(repro): enforce evidence-backed proofs by @zrr1999 in #24
• 🐛 fix(web): hard-cap web-access tool outputs by @zrr1999 in #30
• ✨ feat: add durable repro protocol v4 by @zrr1999 in #28
• 🐛 fix(runtime): preserve session state and retry malformed streams by @zrr1999 in #31
• ✨ feat: establish autonomy correctness contracts by @zrr1999 in #33
• 🐛 fix(ai): decouple stored OAuth reads from kernel registry by @zrr1999 in #34
• 🐛 fix(ai): retry direct provider streams by @zrr1999 in #35
• ♻️ refactor(daemon): converge local RPC on typed oRPC by @zrr1999 in #36
• 🐛 fix(ai): upgrade pi SDK and silence stream logs by @zrr1999 in #38
• ✨ feat(ui): make Cockpit and TUI daemon-first by @zrr1999 in #39
• 🐛 fix(repro): recover daemon drivers across sessions by @zrr1999 in #37
• ✨ feat: simplify Spark learning and command surfaces by @zrr1999 in #40
• ✨ feat(artifacts): persist Product Artifact previews by @zrr1999 in #41
• ✨ feat: add built-in model reproduction skill by @zrr1999 in #44
• ✨ feat(repro): project subgoal orchestration by @zrr1999 in #46
• ✅ test: make driver contracts mechanically auditable by @zrr1999 in #47
• ✅ test: harden side-thread effect isolation by @zrr1999 in #48
• ✨ feat(repro): converge tasks on managed session goals by @zrr1999 in #51
• ✨ feat(tasks): add resource-aware execution scheduling by @zrr1999 in #52
• ✨ feat(repro): add specialized model-reproduction workflows by @zrr1999 in #53
• ✨ feat(cockpit): add goal-first session workbench by @zrr1999 in #45
• 🐛 fix(tui): truncate tool calls by visible width by @zrr1999 in #59
• ✨ feat(artifacts): hard-cut evidence namespace by @zrr1999 in #60
• ✨ feat(memory): refine recall lifecycle and streamline CI by @zrr1999 in #58
• 🐛 fix(ask): persist answer provenance by @zrr1999 in #56
• 🐛 fix(ai): harden Baidu OneAPI prompts and streams by @zrr1999 in #64
• ✨ feat(tasks): add non-terminal claim release by @zrr1999 in #65
• 🐛 fix(evidence): wire canonical ledger tool by @zrr1999 in #25
• 🐛 fix(repro): tolerate legacy completions and retry stream reads by @zrr1999 in #66
• 🐛 fix(cockpit): reuse the registered runtime workspace by @zrr1999 in #70
• 🐛 fix(cli): make nested help read-only and errors actionable by @zrr1999 in #71
• 📝 docs: add complete Spark operator handbook by @zrr1999 in #72
• 🐛 fix(ai): harden Pi compatibility loader by @zrr1999 in #69
• ✨ feat: daemonize Pi and TUI session heartbeats by @zrr1999 in #49
• ♻️ refactor(tasks): move main claim authority to daemon by @zrr1999 in #55
• ♻️ refactor(architecture): eliminate production cycles by @zrr1999 in #61
• 🐛 fix(evidence): canonicalize migrated state fields by @zrr1999 in #68
• 🐛 fix(session): deliver completion notifications reliably by @zrr1999 in #50
• ✨ feat(session): enforce workspace-owned sessions by @zrr1999 in #80
• 🐛 fix(tasks): fence managed task sessions by @zrr1999 in #81
• ✨ feat(cockpit): retire legacy session shell by @zrr1999 in #62
• ✨ feat(tui): prioritize conversation and ask visibility by @zrr1999 in #63
• 👷 ci: unblock trusted Renovate updates by @zrr1999 in #75
• ⬆️ deps: Update dependencies (non-major) by @renovate[bot] in #26
• ✨ feat(update): support package-manager auto-updates by @zrr1999 in #94
• ✨ feat(container): publish Cockpit image to GHCR by @zrr1999 in #95
• ✨ feat(tui): converge native CLI auth and terminal usability by @zrr1999 in #77
• ✅ test: complete Phase 5 test ownership migration by @zrr1999 in #79
• ✅ test(tui): replace Pi loader alias regression by @zrr1999 in #82
• 🐛 fix(daemon): bound SQLite memory and streamed history by @zrr1999 in #83
• ✨ feat(daemon): resume tool calls across restart by @zrr1999 in #78

New Contributors

• @zrr1999 made their first contribution in #2
• @renovate[bot] made their first contribution in #26

Full Changelog: https://github.com/zendev-lab/spark/commits/v0.2.0