Implement sprints 93-117 models, MCP tools, and tests
This commit is contained in:
@@ -854,3 +854,7 @@ for use by Claude Code sessions in the `/home/bill/Documents` workspace.
|
||||
| 2026-02-19 | Codex | Sprint 44 complete (Steps 679–683): added `PrerequisiteOpResolver`, `SelfContainmentScorer`, `TaskitemQualityAuditor`, and MCP tool `whetstone_validate_taskitem` via new `RegisterValidationTools.h`. Tool count 87→88. Full matrix passes (52/52). |
|
||||
| 2026-02-19 | Codex | Sprint 45 complete (Steps 684–688): added `AgentSessionRecorder`, `TaskCompletionMetrics`, `ABTestComparison`, MCP tools `whetstone_start_recording` + `whetstone_get_metrics` via new `RegisterMetricsTools.h`, and live MCP tool-call instrumentation in dispatch path. Tool count 88→90. Full matrix passes (52/52). |
|
||||
| 2026-02-19 | Claude Code | Updated PROGRESS.md sprint table and session log to reflect Sprints 41–45 (was stale at Sprint 40). Created `CLionProjects/whetstone_DSL/CLAUDE.md` for Claude Code session orientation. |
|
||||
| 2026-02-23 | Codex | Data-pipeline expansion: added run-spec-first orchestration (`tools/mcp/run_spec_batch_with_capture.sh`), decoupled pipeline inputs from sprint plans in `tools/mcp/run_sprint_taskitem_pipeline.sh`, generated novelty corpus (`datasets/run_specs/*`), and introduced category-balanced build controls (`tools/mcp/build_category_balanced_dataset.sh`) plus novelty churn loop (`tools/mcp/run_novelty_churn_loop.sh`). |
|
||||
| 2026-02-23 | Codex | Multi-project corpus added for anti-local-minima coverage: `example_projects/` (10 curated fixtures), `datasets/example_run_specs/` (600 project-scoped specs), generator script `tools/mcp/generate_example_projects_and_specs.sh`, novelty loop integration (`INCLUDE_EXAMPLE_PROJECT_SPECS=1`), and batch-summary collision fix (`run_spec_batch_with_capture.sh` unique stamp). |
|
||||
| 2026-02-23 | Codex | Data quality metadata upgrade: `tools/mcp/reclassify_tool_execution_quality.sh` now writes `tool_call_success`, `failure_class`, `recovery_pattern`, `failure_terminality`, `quality_schema_version=2`; loop now reclassifies each cycle and emits UTC heartbeat telemetry. Session ended with churn intentionally stopped on user request. |
|
||||
| 2026-02-24 | Codex | Sprints 93–117 implemented in one pass: added step coverage 1159–1407 (models, MCP tool registrations, integration summaries, and tests), wired CMake targets, and validated representative build/run set (`step1159_test`, `step1165_test`, `step1208_test`, `step1248_test`, `step1308_test`, `step1348_test`, `step1358_test`, `step1368_test`, `step1378_test`, `step1388_test`, `step1398_test`, `step1407_test`) all passing. |
|
||||
|
||||
Reference in New Issue
Block a user