tasks: - name: comet-native-clarification category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Interactive Native workflow that must clarify one high-impact requirement before confirmed execution. - name: comet-native-clarification-depth category: comet-native default_treatments: - COMET_NATIVE_SEQUENTIAL description: Sequential Native benchmark for discovering hidden user-owned decisions before confirmed execution. - name: comet-native-clarification-modes category: comet-native default_treatments: - COMET_NATIVE_SEQUENTIAL - COMET_NATIVE_BATCH description: Paired Native clarification benchmark with three independent decisions for Sequential versus Batch comparison. - name: comet-native-workflow category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Self-contained Shape-Build-Verify-Archive workflow using only the Comet Native Skill and its bundled runtime. - name: comet-native-repository-fact category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native workflow that must investigate a repository-owned product rule instead of asking the user. - name: comet-native-interrupted-transition category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native workflow that must recover a prepared phase transition before implementing and archiving. - name: comet-native-wave-b-decision-resume category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native beta17 task for one product decision and a context-cleared v4 resume from disk. - name: comet-native-wave-c-verification-integrity category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native beta17 task for complete Verifier coverage, repair loop, and no Archive recheck. - name: comet-native-wave-d-stagnation-stop category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native beta17 task for bounded no-progress and execution-error loop stopping. - name: comet-native-wave-e-parallel-safety category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native beta17 task for Runtime lock, stateVersion, and single-winner concurrent mutation. - name: comet-native-wave-f-dashboard-readonly category: comet-native default_treatments: - COMET_NATIVE_PHASE1 description: Native beta17 task for Dashboard v2 projection parity and read-only collection. # ── Default / canonical full-workflow task ────────────────────────────── - name: comet-full-workflow category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Full open-design-build-verify-archive workflow on a sentence-counting feature. The canonical 5-phase task. - name: comet-classic-layout-lifecycle category: comet default_treatments: - COMET_CLASSIC_DOCS_LAYOUT - COMET_CLASSIC_LEGACY_LAYOUT description: Current-checkout Classic lifecycle across the docs and explicit legacy artifact layouts. # ── Dimension coverage tasks (bug fix / refactor / perf / api / robustness) ── - name: comet-fix-median category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Bug-fix dimension. Fix median() for even-length lists in a stats library; diagnose, fix, verify no regressions. - name: comet-refactor-counter category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Refactor dimension. Collapse three duplicated count_* functions into a count(text, unit) dispatcher while preserving behavior. - name: comet-perf-dedupe category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Performance dimension. Optimize an O(n^2) dedupe function to O(n) using a set while preserving first-occurrence order. - name: comet-api-cache-ttl category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: API-design dimension. Add optional TTL expiration to an in-memory cache while keeping the existing API backward compatible. - name: comet-robust-config category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Robustness dimension. Add structured ConfigError handling to a brittle INI config loader so malformed input raises clear errors. # ── Expanded workflow pressure tasks inspired by skills-benchmarks ────── - name: comet-dependency-confusion category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/lc-deps-tavily. Repair stale module names and compatibility imports in a small settings loader without changing the public API. - name: comet-layered-streaming-fix category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/oss-fix-lc-streaming. Fix streaming chunk behavior and tool-result handling in a chat application. - name: comet-persistence-threading category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/oss-fix-lg-persistence. Make session state isolate by thread_id and survive a new store instance. - name: comet-human-approval-flow category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/oss-fix-lc-hitl. Add a pending approval flow before dangerous document operations can execute. - name: comet-noise-distractor category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/lc-basic-noise. Fix invoice tax calculation while leaving an unrelated distracting file untouched. - name: comet-cross-file-refactor category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/lc-basic. Add a shared count(text, unit) dispatcher while keeping wrappers and CLI behavior compatible. - name: comet-observability-env-template category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/ls-lang-tracing. Generate a LangSmith-ready .env.example with the simplified LANGSMITH_* configuration. - name: comet-graph-execution-review category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/oss-fix-lg-execution. Fix parallel task fan-out, review interruption, and resume behavior in a graph-style pipeline. - name: comet-agent-memory-routing category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/oss-fix-da-memory. Fix persistent preference routing, subagent skill access, and deployment approval checkpoints. - name: comet-framework-selection category: comet default_treatments: - CONTROL - COMET_FULL_040_BETA - COMET_FULL_039 description: Inspired by skills-benchmarks/lc-framework-hybrid-pipeline. Select a hybrid deep-agent plus deterministic sub-workflow architecture for long-session document editing. - name: generic-skill-smoke category: generic default_treatments: - CONTROL description: Generic smoke task for arbitrary Skill eval; creates and validates result.md. - name: authoring-skill-smoke category: generic default_treatments: - CONTROL description: Smoke task for generated Comet-native Skill packages. - name: workflow-route-conformance category: generic default_treatments: - CONTROL description: Validates generated Skill route metadata against the workflow protocol and internal Node Skills. - name: workflow-overlay-contract category: generic default_treatments: - CONTROL description: Validates generated overlay eval manifests for baselines, gates, schemas, and expected evidence.