cognitive-core 0.2.4 → 0.3.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +21 -0
- package/README.md +88 -143
- package/dist/agents/mock-provider.d.ts.map +1 -1
- package/dist/agents/mock-provider.js.map +1 -1
- package/dist/agents/types.d.ts.map +1 -1
- package/dist/agents/types.js +5 -4
- package/dist/agents/types.js.map +1 -1
- package/dist/atlas.d.ts +309 -97
- package/dist/atlas.d.ts.map +1 -1
- package/dist/atlas.js +743 -277
- package/dist/atlas.js.map +1 -1
- package/dist/bin/cli-utils.d.ts +25 -2
- package/dist/bin/cli-utils.d.ts.map +1 -1
- package/dist/bin/cli-utils.js +79 -9
- package/dist/bin/cli-utils.js.map +1 -1
- package/dist/bin/cognitive-core.js +32 -0
- package/dist/bin/cognitive-core.js.map +1 -1
- package/dist/bin/commands/dream.d.ts +17 -0
- package/dist/bin/commands/dream.d.ts.map +1 -0
- package/dist/bin/commands/dream.js +168 -0
- package/dist/bin/commands/dream.js.map +1 -0
- package/dist/bin/commands/eval.d.ts +181 -0
- package/dist/bin/commands/eval.d.ts.map +1 -0
- package/dist/bin/commands/eval.js +1582 -0
- package/dist/bin/commands/eval.js.map +1 -0
- package/dist/bin/commands/kb.js +0 -8
- package/dist/bin/commands/kb.js.map +1 -1
- package/dist/bin/commands/learn.js +12 -4
- package/dist/bin/commands/learn.js.map +1 -1
- package/dist/bin/commands/legacy.d.ts.map +1 -1
- package/dist/bin/commands/legacy.js +1 -1
- package/dist/bin/commands/legacy.js.map +1 -1
- package/dist/bin/commands/run.d.ts.map +1 -1
- package/dist/bin/commands/run.js +10 -9
- package/dist/bin/commands/run.js.map +1 -1
- package/dist/bin/commands/sessions.d.ts.map +1 -1
- package/dist/bin/commands/sessions.js.map +1 -1
- package/dist/bin/commands/skills.d.ts.map +1 -1
- package/dist/bin/commands/skills.js +283 -1
- package/dist/bin/commands/skills.js.map +1 -1
- package/dist/embeddings/bm25.d.ts.map +1 -1
- package/dist/embeddings/bm25.js +1 -2
- package/dist/embeddings/bm25.js.map +1 -1
- package/dist/embeddings/index.d.ts +3 -3
- package/dist/embeddings/index.d.ts.map +1 -1
- package/dist/embeddings/index.js +3 -3
- package/dist/embeddings/index.js.map +1 -1
- package/dist/embeddings/inverted-index.js +62 -7
- package/dist/embeddings/inverted-index.js.map +1 -1
- package/dist/embeddings/manager.d.ts.map +1 -1
- package/dist/embeddings/manager.js +6 -5
- package/dist/embeddings/manager.js.map +1 -1
- package/dist/embeddings/provider.d.ts +96 -5
- package/dist/embeddings/provider.d.ts.map +1 -1
- package/dist/embeddings/provider.js +171 -48
- package/dist/embeddings/provider.js.map +1 -1
- package/dist/embeddings/vector-store.d.ts.map +1 -1
- package/dist/embeddings/vector-store.js +3 -1
- package/dist/embeddings/vector-store.js.map +1 -1
- package/dist/eval/ablation-validate.d.ts +12 -0
- package/dist/eval/ablation-validate.d.ts.map +1 -0
- package/dist/eval/ablation-validate.js +78 -0
- package/dist/eval/ablation-validate.js.map +1 -0
- package/dist/eval/ablation.d.ts +94 -0
- package/dist/eval/ablation.d.ts.map +1 -0
- package/dist/eval/ablation.js +161 -0
- package/dist/eval/ablation.js.map +1 -0
- package/dist/eval/ci-smoke.d.ts +11 -0
- package/dist/eval/ci-smoke.d.ts.map +1 -0
- package/dist/eval/ci-smoke.js +49 -0
- package/dist/eval/ci-smoke.js.map +1 -0
- package/dist/eval/diagnostics.d.ts +214 -0
- package/dist/eval/diagnostics.d.ts.map +1 -0
- package/dist/eval/diagnostics.js +107 -0
- package/dist/eval/diagnostics.js.map +1 -0
- package/dist/eval/index.d.ts +27 -0
- package/dist/eval/index.d.ts.map +1 -0
- package/dist/eval/index.js +27 -0
- package/dist/eval/index.js.map +1 -0
- package/dist/eval/live/alfworld.d.ts +248 -0
- package/dist/eval/live/alfworld.d.ts.map +1 -0
- package/dist/eval/live/alfworld.js +637 -0
- package/dist/eval/live/alfworld.js.map +1 -0
- package/dist/eval/live/alfworld_h1.d.ts +16 -0
- package/dist/eval/live/alfworld_h1.d.ts.map +1 -0
- package/dist/eval/live/alfworld_h1.js +75 -0
- package/dist/eval/live/alfworld_h1.js.map +1 -0
- package/dist/eval/live/alfworld_memory.d.ts +38 -0
- package/dist/eval/live/alfworld_memory.d.ts.map +1 -0
- package/dist/eval/live/alfworld_memory.js +184 -0
- package/dist/eval/live/alfworld_memory.js.map +1 -0
- package/dist/eval/live/alfworld_strategy.d.ts +65 -0
- package/dist/eval/live/alfworld_strategy.d.ts.map +1 -0
- package/dist/eval/live/alfworld_strategy.js +89 -0
- package/dist/eval/live/alfworld_strategy.js.map +1 -0
- package/dist/eval/live/azure-openai-llm-provider.d.ts +41 -0
- package/dist/eval/live/azure-openai-llm-provider.d.ts.map +1 -0
- package/dist/eval/live/azure-openai-llm-provider.js +92 -0
- package/dist/eval/live/azure-openai-llm-provider.js.map +1 -0
- package/dist/eval/live/bridges/skillflow_e2b_bridge.d.ts +3 -0
- package/dist/eval/live/bridges/skillflow_e2b_bridge.d.ts.map +1 -0
- package/dist/eval/live/bridges/skillflow_e2b_bridge.js +12 -0
- package/dist/eval/live/bridges/skillflow_e2b_bridge.js.map +1 -0
- package/dist/eval/live/claude-cli-llm-provider.d.ts +53 -0
- package/dist/eval/live/claude-cli-llm-provider.d.ts.map +1 -0
- package/dist/eval/live/claude-cli-llm-provider.js +71 -0
- package/dist/eval/live/claude-cli-llm-provider.js.map +1 -0
- package/dist/eval/live/claude-cli.d.ts +96 -0
- package/dist/eval/live/claude-cli.d.ts.map +1 -0
- package/dist/eval/live/claude-cli.js +314 -0
- package/dist/eval/live/claude-cli.js.map +1 -0
- package/dist/eval/live/claude-delegate.d.ts +41 -0
- package/dist/eval/live/claude-delegate.d.ts.map +1 -0
- package/dist/eval/live/claude-delegate.js +59 -0
- package/dist/eval/live/claude-delegate.js.map +1 -0
- package/dist/eval/live/claude-workspace-backend.d.ts +24 -0
- package/dist/eval/live/claude-workspace-backend.d.ts.map +1 -0
- package/dist/eval/live/claude-workspace-backend.js +76 -0
- package/dist/eval/live/claude-workspace-backend.js.map +1 -0
- package/dist/eval/live/clbench.d.ts +64 -0
- package/dist/eval/live/clbench.d.ts.map +1 -0
- package/dist/eval/live/clbench.js +291 -0
- package/dist/eval/live/clbench.js.map +1 -0
- package/dist/eval/live/cold-reuse.d.ts +21 -0
- package/dist/eval/live/cold-reuse.d.ts.map +1 -0
- package/dist/eval/live/cold-reuse.js +102 -0
- package/dist/eval/live/cold-reuse.js.map +1 -0
- package/dist/eval/live/difficulty-registry.d.ts +68 -0
- package/dist/eval/live/difficulty-registry.d.ts.map +1 -0
- package/dist/eval/live/difficulty-registry.js +169 -0
- package/dist/eval/live/difficulty-registry.js.map +1 -0
- package/dist/eval/live/docker.d.ts +52 -0
- package/dist/eval/live/docker.d.ts.map +1 -0
- package/dist/eval/live/docker.js +164 -0
- package/dist/eval/live/docker.js.map +1 -0
- package/dist/eval/live/env-bridge.d.ts +68 -0
- package/dist/eval/live/env-bridge.d.ts.map +1 -0
- package/dist/eval/live/env-bridge.js +188 -0
- package/dist/eval/live/env-bridge.js.map +1 -0
- package/dist/eval/live/gsm8k.d.ts +44 -0
- package/dist/eval/live/gsm8k.d.ts.map +1 -0
- package/dist/eval/live/gsm8k.js +142 -0
- package/dist/eval/live/gsm8k.js.map +1 -0
- package/dist/eval/live/index.d.ts +38 -0
- package/dist/eval/live/index.d.ts.map +1 -0
- package/dist/eval/live/index.js +38 -0
- package/dist/eval/live/index.js.map +1 -0
- package/dist/eval/live/lifelongagent.d.ts +64 -0
- package/dist/eval/live/lifelongagent.d.ts.map +1 -0
- package/dist/eval/live/lifelongagent.js +216 -0
- package/dist/eval/live/lifelongagent.js.map +1 -0
- package/dist/eval/live/lift-registry.d.ts +128 -0
- package/dist/eval/live/lift-registry.d.ts.map +1 -0
- package/dist/eval/live/lift-registry.js +147 -0
- package/dist/eval/live/lift-registry.js.map +1 -0
- package/dist/eval/live/live-benchmark.d.ts +84 -0
- package/dist/eval/live/live-benchmark.d.ts.map +1 -0
- package/dist/eval/live/live-benchmark.js +2 -0
- package/dist/eval/live/live-benchmark.js.map +1 -0
- package/dist/eval/live/live-runner.d.ts +609 -0
- package/dist/eval/live/live-runner.d.ts.map +1 -0
- package/dist/eval/live/live-runner.js +1409 -0
- package/dist/eval/live/live-runner.js.map +1 -0
- package/dist/eval/live/refine.d.ts +88 -0
- package/dist/eval/live/refine.d.ts.map +1 -0
- package/dist/eval/live/refine.js +135 -0
- package/dist/eval/live/refine.js.map +1 -0
- package/dist/eval/live/registry-store.d.ts +109 -0
- package/dist/eval/live/registry-store.d.ts.map +1 -0
- package/dist/eval/live/registry-store.js +272 -0
- package/dist/eval/live/registry-store.js.map +1 -0
- package/dist/eval/live/run-checkpoint.d.ts +55 -0
- package/dist/eval/live/run-checkpoint.d.ts.map +1 -0
- package/dist/eval/live/run-checkpoint.js +86 -0
- package/dist/eval/live/run-checkpoint.js.map +1 -0
- package/dist/eval/live/run-resume.d.ts +22 -0
- package/dist/eval/live/run-resume.d.ts.map +1 -0
- package/dist/eval/live/run-resume.js +70 -0
- package/dist/eval/live/run-resume.js.map +1 -0
- package/dist/eval/live/skill-injector.d.ts +36 -0
- package/dist/eval/live/skill-injector.d.ts.map +1 -0
- package/dist/eval/live/skill-injector.js +81 -0
- package/dist/eval/live/skill-injector.js.map +1 -0
- package/dist/eval/live/skillflow-embedded-workbook-bundle.d.ts +3 -0
- package/dist/eval/live/skillflow-embedded-workbook-bundle.d.ts.map +1 -0
- package/dist/eval/live/skillflow-embedded-workbook-bundle.js +391 -0
- package/dist/eval/live/skillflow-embedded-workbook-bundle.js.map +1 -0
- package/dist/eval/live/skillflow.d.ts +154 -0
- package/dist/eval/live/skillflow.d.ts.map +1 -0
- package/dist/eval/live/skillflow.js +612 -0
- package/dist/eval/live/skillflow.js.map +1 -0
- package/dist/eval/live/skilllearnbench.d.ts +43 -0
- package/dist/eval/live/skilllearnbench.d.ts.map +1 -0
- package/dist/eval/live/skilllearnbench.js +154 -0
- package/dist/eval/live/skilllearnbench.js.map +1 -0
- package/dist/eval/live/skillopt_v1.d.ts +96 -0
- package/dist/eval/live/skillopt_v1.d.ts.map +1 -0
- package/dist/eval/live/skillopt_v1.js +0 -0
- package/dist/eval/live/skillopt_v1.js.map +1 -0
- package/dist/eval/live/skillopt_v1_causal.d.ts +26 -0
- package/dist/eval/live/skillopt_v1_causal.d.ts.map +1 -0
- package/dist/eval/live/skillopt_v1_causal.js +106 -0
- package/dist/eval/live/skillopt_v1_causal.js.map +1 -0
- package/dist/eval/live/skillopt_v1_memory.d.ts +53 -0
- package/dist/eval/live/skillopt_v1_memory.d.ts.map +1 -0
- package/dist/eval/live/skillopt_v1_memory.js +147 -0
- package/dist/eval/live/skillopt_v1_memory.js.map +1 -0
- package/dist/eval/live/skillopt_v1_rules.d.ts +39 -0
- package/dist/eval/live/skillopt_v1_rules.d.ts.map +1 -0
- package/dist/eval/live/skillopt_v1_rules.js +91 -0
- package/dist/eval/live/skillopt_v1_rules.js.map +1 -0
- package/dist/eval/live/skillsbench.d.ts +138 -0
- package/dist/eval/live/skillsbench.d.ts.map +1 -0
- package/dist/eval/live/skillsbench.js +626 -0
- package/dist/eval/live/skillsbench.js.map +1 -0
- package/dist/eval/live/stopping.d.ts +69 -0
- package/dist/eval/live/stopping.d.ts.map +1 -0
- package/dist/eval/live/stopping.js +213 -0
- package/dist/eval/live/stopping.js.map +1 -0
- package/dist/eval/live/trace.d.ts +20 -0
- package/dist/eval/live/trace.d.ts.map +1 -0
- package/dist/eval/live/trace.js +41 -0
- package/dist/eval/live/trace.js.map +1 -0
- package/dist/eval/mechanism-trace-adapter.d.ts +8 -0
- package/dist/eval/mechanism-trace-adapter.d.ts.map +1 -0
- package/dist/eval/mechanism-trace-adapter.js +118 -0
- package/dist/eval/mechanism-trace-adapter.js.map +1 -0
- package/dist/eval/memory-ab.d.ts +61 -0
- package/dist/eval/memory-ab.d.ts.map +1 -0
- package/dist/eval/memory-ab.js +161 -0
- package/dist/eval/memory-ab.js.map +1 -0
- package/dist/eval/memory-mock.d.ts +54 -0
- package/dist/eval/memory-mock.d.ts.map +1 -0
- package/dist/eval/memory-mock.js +129 -0
- package/dist/eval/memory-mock.js.map +1 -0
- package/dist/eval/memory-qa/index.d.ts +13 -0
- package/dist/eval/memory-qa/index.d.ts.map +1 -0
- package/dist/eval/memory-qa/index.js +13 -0
- package/dist/eval/memory-qa/index.js.map +1 -0
- package/dist/eval/memory-qa/iterative.d.ts +80 -0
- package/dist/eval/memory-qa/iterative.d.ts.map +1 -0
- package/dist/eval/memory-qa/iterative.js +107 -0
- package/dist/eval/memory-qa/iterative.js.map +1 -0
- package/dist/eval/memory-qa/loaders.d.ts +11 -0
- package/dist/eval/memory-qa/loaders.d.ts.map +1 -0
- package/dist/eval/memory-qa/loaders.js +21 -0
- package/dist/eval/memory-qa/loaders.js.map +1 -0
- package/dist/eval/memory-qa/metrics.d.ts +3 -0
- package/dist/eval/memory-qa/metrics.d.ts.map +1 -0
- package/dist/eval/memory-qa/metrics.js +2 -0
- package/dist/eval/memory-qa/metrics.js.map +1 -0
- package/dist/eval/memory-qa/qa.d.ts +186 -0
- package/dist/eval/memory-qa/qa.d.ts.map +1 -0
- package/dist/eval/memory-qa/qa.js +424 -0
- package/dist/eval/memory-qa/qa.js.map +1 -0
- package/dist/eval/memory-qa/retrieval.d.ts +33 -0
- package/dist/eval/memory-qa/retrieval.d.ts.map +1 -0
- package/dist/eval/memory-qa/retrieval.js +44 -0
- package/dist/eval/memory-qa/retrieval.js.map +1 -0
- package/dist/eval/memory-qa/serving.d.ts +60 -0
- package/dist/eval/memory-qa/serving.d.ts.map +1 -0
- package/dist/eval/memory-qa/serving.js +267 -0
- package/dist/eval/memory-qa/serving.js.map +1 -0
- package/dist/eval/memory-qa/types.d.ts +2 -0
- package/dist/eval/memory-qa/types.d.ts.map +1 -0
- package/dist/eval/memory-qa/types.js +2 -0
- package/dist/eval/memory-qa/types.js.map +1 -0
- package/dist/eval/metrics.d.ts +120 -0
- package/dist/eval/metrics.d.ts.map +1 -0
- package/dist/eval/metrics.js +361 -0
- package/dist/eval/metrics.js.map +1 -0
- package/dist/eval/oracle-smoke.d.ts +75 -0
- package/dist/eval/oracle-smoke.d.ts.map +1 -0
- package/dist/eval/oracle-smoke.js +143 -0
- package/dist/eval/oracle-smoke.js.map +1 -0
- package/dist/eval/playbook-calibration-report.d.ts +40 -0
- package/dist/eval/playbook-calibration-report.d.ts.map +1 -0
- package/dist/eval/playbook-calibration-report.js +246 -0
- package/dist/eval/playbook-calibration-report.js.map +1 -0
- package/dist/eval/playbook-candidate-decisions-report.d.ts +36 -0
- package/dist/eval/playbook-candidate-decisions-report.d.ts.map +1 -0
- package/dist/eval/playbook-candidate-decisions-report.js +201 -0
- package/dist/eval/playbook-candidate-decisions-report.js.map +1 -0
- package/dist/eval/playbook-usage-report.d.ts +7 -0
- package/dist/eval/playbook-usage-report.d.ts.map +1 -0
- package/dist/eval/playbook-usage-report.js +222 -0
- package/dist/eval/playbook-usage-report.js.map +1 -0
- package/dist/eval/report.d.ts +28 -0
- package/dist/eval/report.d.ts.map +1 -0
- package/dist/eval/report.js +567 -0
- package/dist/eval/report.js.map +1 -0
- package/dist/eval/retrieval/corpus.d.ts +41 -0
- package/dist/eval/retrieval/corpus.d.ts.map +1 -0
- package/dist/eval/retrieval/corpus.js +158 -0
- package/dist/eval/retrieval/corpus.js.map +1 -0
- package/dist/eval/retrieval/index.d.ts +12 -0
- package/dist/eval/retrieval/index.d.ts.map +1 -0
- package/dist/eval/retrieval/index.js +14 -0
- package/dist/eval/retrieval/index.js.map +1 -0
- package/dist/eval/retrieval/knowledge-corpus.d.ts +91 -0
- package/dist/eval/retrieval/knowledge-corpus.d.ts.map +1 -0
- package/dist/eval/retrieval/knowledge-corpus.js +268 -0
- package/dist/eval/retrieval/knowledge-corpus.js.map +1 -0
- package/dist/eval/retrieval/knowledge-quality.d.ts +91 -0
- package/dist/eval/retrieval/knowledge-quality.d.ts.map +1 -0
- package/dist/eval/retrieval/knowledge-quality.js +247 -0
- package/dist/eval/retrieval/knowledge-quality.js.map +1 -0
- package/dist/eval/retrieval/ladder.d.ts +69 -0
- package/dist/eval/retrieval/ladder.d.ts.map +1 -0
- package/dist/eval/retrieval/ladder.js +80 -0
- package/dist/eval/retrieval/ladder.js.map +1 -0
- package/dist/eval/retrieval/quality.d.ts +39 -0
- package/dist/eval/retrieval/quality.d.ts.map +1 -0
- package/dist/eval/retrieval/quality.js +75 -0
- package/dist/eval/retrieval/quality.js.map +1 -0
- package/dist/eval/rng.d.ts +25 -0
- package/dist/eval/rng.d.ts.map +1 -0
- package/dist/eval/rng.js +43 -0
- package/dist/eval/rng.js.map +1 -0
- package/dist/eval/runner.d.ts +65 -0
- package/dist/eval/runner.d.ts.map +1 -0
- package/dist/eval/runner.js +216 -0
- package/dist/eval/runner.js.map +1 -0
- package/dist/eval/skill-bundle.d.ts +18 -0
- package/dist/eval/skill-bundle.d.ts.map +1 -0
- package/dist/eval/skill-bundle.js +38 -0
- package/dist/eval/skill-bundle.js.map +1 -0
- package/dist/eval/swarmkit-report.d.ts +21 -0
- package/dist/eval/swarmkit-report.d.ts.map +1 -0
- package/dist/eval/swarmkit-report.js +93 -0
- package/dist/eval/swarmkit-report.js.map +1 -0
- package/dist/eval/task-stream.d.ts +22 -0
- package/dist/eval/task-stream.d.ts.map +1 -0
- package/dist/eval/task-stream.js +99 -0
- package/dist/eval/task-stream.js.map +1 -0
- package/dist/eval/types.d.ts +173 -0
- package/dist/eval/types.d.ts.map +1 -0
- package/dist/eval/types.js +11 -0
- package/dist/eval/types.js.map +1 -0
- package/dist/experimental/causal-store.d.ts +97 -0
- package/dist/experimental/causal-store.d.ts.map +1 -0
- package/dist/experimental/causal-store.js +213 -0
- package/dist/experimental/causal-store.js.map +1 -0
- package/dist/experimental/graph-layers/base.d.ts +29 -0
- package/dist/experimental/graph-layers/base.d.ts.map +1 -0
- package/dist/experimental/graph-layers/base.js +143 -0
- package/dist/experimental/graph-layers/base.js.map +1 -0
- package/dist/experimental/graph-layers/causal.d.ts +14 -0
- package/dist/experimental/graph-layers/causal.d.ts.map +1 -0
- package/dist/experimental/graph-layers/causal.js +14 -0
- package/dist/experimental/graph-layers/causal.js.map +1 -0
- package/dist/experimental/graph-layers/entity.d.ts +14 -0
- package/dist/experimental/graph-layers/entity.d.ts.map +1 -0
- package/dist/experimental/graph-layers/entity.js +14 -0
- package/dist/experimental/graph-layers/entity.js.map +1 -0
- package/dist/experimental/graph-layers/index.d.ts +6 -0
- package/dist/experimental/graph-layers/index.d.ts.map +1 -0
- package/dist/experimental/graph-layers/index.js +6 -0
- package/dist/experimental/graph-layers/index.js.map +1 -0
- package/dist/experimental/graph-layers/semantic.d.ts +14 -0
- package/dist/experimental/graph-layers/semantic.d.ts.map +1 -0
- package/dist/experimental/graph-layers/semantic.js +14 -0
- package/dist/experimental/graph-layers/semantic.js.map +1 -0
- package/dist/experimental/graph-layers/temporal.d.ts +14 -0
- package/dist/experimental/graph-layers/temporal.d.ts.map +1 -0
- package/dist/experimental/graph-layers/temporal.js +14 -0
- package/dist/experimental/graph-layers/temporal.js.map +1 -0
- package/dist/experimental/healing-orchestrator.d.ts +182 -0
- package/dist/experimental/healing-orchestrator.d.ts.map +1 -0
- package/dist/experimental/healing-orchestrator.js +250 -0
- package/dist/experimental/healing-orchestrator.js.map +1 -0
- package/dist/experimental/knowledge-graph-types.d.ts +148 -0
- package/dist/experimental/knowledge-graph-types.d.ts.map +1 -0
- package/dist/experimental/knowledge-graph-types.js +40 -0
- package/dist/experimental/knowledge-graph-types.js.map +1 -0
- package/dist/experimental/knowledge-graph.d.ts +41 -0
- package/dist/experimental/knowledge-graph.d.ts.map +1 -0
- package/dist/experimental/knowledge-graph.js +270 -0
- package/dist/experimental/knowledge-graph.js.map +1 -0
- package/dist/experimental/meta-reflection.d.ts +47 -0
- package/dist/experimental/meta-reflection.d.ts.map +1 -0
- package/dist/experimental/meta-reflection.js +143 -0
- package/dist/experimental/meta-reflection.js.map +1 -0
- package/dist/experimental/moe-gate.d.ts +134 -0
- package/dist/experimental/moe-gate.d.ts.map +1 -0
- package/dist/experimental/moe-gate.js +240 -0
- package/dist/experimental/moe-gate.js.map +1 -0
- package/dist/experimental/reasoning-bank.d.ts +130 -0
- package/dist/experimental/reasoning-bank.d.ts.map +1 -0
- package/dist/experimental/reasoning-bank.js +427 -0
- package/dist/experimental/reasoning-bank.js.map +1 -0
- package/dist/experimental/team/team-config.d.ts +127 -0
- package/dist/experimental/team/team-config.d.ts.map +1 -0
- package/dist/experimental/team/team-config.js +48 -0
- package/dist/experimental/team/team-config.js.map +1 -0
- package/dist/experimental/team/team-experience.d.ts +300 -0
- package/dist/experimental/team/team-experience.d.ts.map +1 -0
- package/dist/experimental/team/team-experience.js +355 -0
- package/dist/experimental/team/team-experience.js.map +1 -0
- package/dist/experimental/team/team-ingester.d.ts +152 -0
- package/dist/experimental/team/team-ingester.d.ts.map +1 -0
- package/dist/experimental/team/team-ingester.js +330 -0
- package/dist/experimental/team/team-ingester.js.map +1 -0
- package/dist/experimental/team/team-meta-learner.d.ts +51 -0
- package/dist/experimental/team/team-meta-learner.d.ts.map +1 -0
- package/dist/experimental/team/team-meta-learner.js +404 -0
- package/dist/experimental/team/team-meta-learner.js.map +1 -0
- package/dist/experimental/team/team-meta.d.ts +160 -0
- package/dist/experimental/team/team-meta.d.ts.map +1 -0
- package/dist/experimental/team/team-meta.js +42 -0
- package/dist/experimental/team/team-meta.js.map +1 -0
- package/dist/experimental/team/team-pipeline.d.ts +77 -0
- package/dist/experimental/team/team-pipeline.d.ts.map +1 -0
- package/dist/experimental/team/team-pipeline.js +262 -0
- package/dist/experimental/team/team-pipeline.js.map +1 -0
- package/dist/experimental/team/team-playbook-extraction.d.ts +45 -0
- package/dist/experimental/team/team-playbook-extraction.d.ts.map +1 -0
- package/dist/experimental/team/team-playbook-extraction.js +526 -0
- package/dist/experimental/team/team-playbook-extraction.js.map +1 -0
- package/dist/experimental/team/team-playbook.d.ts +276 -0
- package/dist/experimental/team/team-playbook.d.ts.map +1 -0
- package/dist/experimental/team/team-playbook.js +85 -0
- package/dist/experimental/team/team-playbook.js.map +1 -0
- package/dist/experimental/team/team-router.d.ts +98 -0
- package/dist/experimental/team/team-router.d.ts.map +1 -0
- package/dist/experimental/team/team-router.js +312 -0
- package/dist/experimental/team/team-router.js.map +1 -0
- package/dist/experimental/team/team-skill-library.d.ts +180 -0
- package/dist/experimental/team/team-skill-library.d.ts.map +1 -0
- package/dist/experimental/team/team-skill-library.js +385 -0
- package/dist/experimental/team/team-skill-library.js.map +1 -0
- package/dist/experimental/team/team-trajectory-analysis.d.ts +19 -0
- package/dist/experimental/team/team-trajectory-analysis.d.ts.map +1 -0
- package/dist/experimental/team/team-trajectory-analysis.js +442 -0
- package/dist/experimental/team/team-trajectory-analysis.js.map +1 -0
- package/dist/experimental/team/team-trajectory.d.ts +305 -0
- package/dist/experimental/team/team-trajectory.d.ts.map +1 -0
- package/dist/experimental/team/team-trajectory.js +315 -0
- package/dist/experimental/team/team-trajectory.js.map +1 -0
- package/dist/factory.d.ts +2 -3
- package/dist/factory.d.ts.map +1 -1
- package/dist/factory.js +12 -11
- package/dist/factory.js.map +1 -1
- package/dist/index.d.ts +9 -11
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +26 -32
- package/dist/index.js.map +1 -1
- package/dist/learning/analyzer.d.ts +26 -36
- package/dist/learning/analyzer.d.ts.map +1 -1
- package/dist/learning/analyzer.js +50 -11
- package/dist/learning/analyzer.js.map +1 -1
- package/dist/learning/compound-bootstrap/external-acquisition.d.ts +63 -0
- package/dist/learning/compound-bootstrap/external-acquisition.d.ts.map +1 -0
- package/dist/learning/compound-bootstrap/external-acquisition.js +255 -0
- package/dist/learning/compound-bootstrap/external-acquisition.js.map +1 -0
- package/dist/learning/compound-bootstrap/index.d.ts +3 -0
- package/dist/learning/compound-bootstrap/index.d.ts.map +1 -0
- package/dist/learning/compound-bootstrap/index.js +3 -0
- package/dist/learning/compound-bootstrap/index.js.map +1 -0
- package/dist/learning/compound-bootstrap/workspace-analyzer.d.ts +33 -0
- package/dist/learning/compound-bootstrap/workspace-analyzer.d.ts.map +1 -0
- package/dist/learning/compound-bootstrap/workspace-analyzer.js +379 -0
- package/dist/learning/compound-bootstrap/workspace-analyzer.js.map +1 -0
- package/dist/learning/consolidation-coordinator.d.ts +49 -0
- package/dist/learning/consolidation-coordinator.d.ts.map +1 -0
- package/dist/learning/consolidation-coordinator.js +75 -0
- package/dist/learning/consolidation-coordinator.js.map +1 -0
- package/dist/learning/convention-distiller.d.ts +84 -0
- package/dist/learning/convention-distiller.d.ts.map +1 -0
- package/dist/learning/convention-distiller.js +268 -0
- package/dist/learning/convention-distiller.js.map +1 -0
- package/dist/learning/distiller.d.ts +82 -0
- package/dist/learning/distiller.d.ts.map +1 -0
- package/dist/learning/distiller.js +353 -0
- package/dist/learning/distiller.js.map +1 -0
- package/dist/learning/effectiveness.d.ts +12 -2
- package/dist/learning/effectiveness.d.ts.map +1 -1
- package/dist/learning/effectiveness.js +18 -14
- package/dist/learning/effectiveness.js.map +1 -1
- package/dist/learning/escalation.d.ts +89 -0
- package/dist/learning/escalation.d.ts.map +1 -0
- package/dist/learning/escalation.js +168 -0
- package/dist/learning/escalation.js.map +1 -0
- package/dist/learning/external-exposure.d.ts +57 -0
- package/dist/learning/external-exposure.d.ts.map +1 -0
- package/dist/learning/external-exposure.js +95 -0
- package/dist/learning/external-exposure.js.map +1 -0
- package/dist/learning/gepa-proposer.d.ts +116 -0
- package/dist/learning/gepa-proposer.d.ts.map +1 -0
- package/dist/learning/gepa-proposer.js +258 -0
- package/dist/learning/gepa-proposer.js.map +1 -0
- package/dist/learning/index.d.ts +21 -8
- package/dist/learning/index.d.ts.map +1 -1
- package/dist/learning/index.js +28 -10
- package/dist/learning/index.js.map +1 -1
- package/dist/learning/instant-loop.d.ts +46 -16
- package/dist/learning/instant-loop.d.ts.map +1 -1
- package/dist/learning/instant-loop.js +114 -111
- package/dist/learning/instant-loop.js.map +1 -1
- package/dist/learning/knowledge-extractor.d.ts.map +1 -1
- package/dist/learning/knowledge-extractor.js +1 -2
- package/dist/learning/knowledge-extractor.js.map +1 -1
- package/dist/learning/llm-extractor.d.ts +88 -0
- package/dist/learning/llm-extractor.d.ts.map +1 -0
- package/dist/learning/llm-extractor.js +372 -0
- package/dist/learning/llm-extractor.js.map +1 -0
- package/dist/learning/loop-coordinator.d.ts +61 -0
- package/dist/learning/loop-coordinator.d.ts.map +1 -0
- package/dist/learning/loop-coordinator.js +96 -0
- package/dist/learning/loop-coordinator.js.map +1 -0
- package/dist/learning/maintenance-scheduler.d.ts +15 -4
- package/dist/learning/maintenance-scheduler.d.ts.map +1 -1
- package/dist/learning/maintenance-scheduler.js +19 -16
- package/dist/learning/maintenance-scheduler.js.map +1 -1
- package/dist/learning/mechanism-trace.d.ts +196 -0
- package/dist/learning/mechanism-trace.d.ts.map +1 -0
- package/dist/learning/mechanism-trace.js +81 -0
- package/dist/learning/mechanism-trace.js.map +1 -0
- package/dist/learning/merge-proposer.d.ts +102 -0
- package/dist/learning/merge-proposer.d.ts.map +1 -0
- package/dist/learning/merge-proposer.js +253 -0
- package/dist/learning/merge-proposer.js.map +1 -0
- package/dist/learning/meta-learner.d.ts +9 -7
- package/dist/learning/meta-learner.d.ts.map +1 -1
- package/dist/learning/meta-learner.js +16 -45
- package/dist/learning/meta-learner.js.map +1 -1
- package/dist/learning/mutation-gate.d.ts +218 -0
- package/dist/learning/mutation-gate.d.ts.map +1 -0
- package/dist/learning/mutation-gate.js +390 -0
- package/dist/learning/mutation-gate.js.map +1 -0
- package/dist/learning/oracle-signal.d.ts +67 -0
- package/dist/learning/oracle-signal.d.ts.map +1 -0
- package/dist/learning/oracle-signal.js +94 -0
- package/dist/learning/oracle-signal.js.map +1 -0
- package/dist/learning/pipeline.d.ts +31 -4
- package/dist/learning/pipeline.d.ts.map +1 -1
- package/dist/learning/pipeline.js +64 -12
- package/dist/learning/pipeline.js.map +1 -1
- package/dist/learning/playbook-extractor.d.ts +8 -0
- package/dist/learning/playbook-extractor.d.ts.map +1 -1
- package/dist/learning/playbook-extractor.js +119 -34
- package/dist/learning/playbook-extractor.js.map +1 -1
- package/dist/learning/playbook-optimizer.d.ts +37 -0
- package/dist/learning/playbook-optimizer.d.ts.map +1 -0
- package/dist/learning/playbook-optimizer.js +125 -0
- package/dist/learning/playbook-optimizer.js.map +1 -0
- package/dist/learning/reflective-proposer.d.ts +173 -0
- package/dist/learning/reflective-proposer.d.ts.map +1 -0
- package/dist/learning/reflective-proposer.js +354 -0
- package/dist/learning/reflective-proposer.js.map +1 -0
- package/dist/learning/reflexion-generator.d.ts.map +1 -1
- package/dist/learning/reflexion-generator.js +4 -3
- package/dist/learning/reflexion-generator.js.map +1 -1
- package/dist/learning/replay-executor.d.ts +64 -0
- package/dist/learning/replay-executor.d.ts.map +1 -0
- package/dist/learning/replay-executor.js +92 -0
- package/dist/learning/replay-executor.js.map +1 -0
- package/dist/learning/replay-types.d.ts +150 -0
- package/dist/learning/replay-types.d.ts.map +1 -0
- package/dist/learning/replay-types.js +207 -0
- package/dist/learning/replay-types.js.map +1 -0
- package/dist/learning/replay.d.ts +166 -0
- package/dist/learning/replay.d.ts.map +1 -0
- package/dist/learning/replay.js +228 -0
- package/dist/learning/replay.js.map +1 -0
- package/dist/learning/seed-report.d.ts +88 -0
- package/dist/learning/seed-report.d.ts.map +1 -0
- package/dist/learning/seed-report.js +120 -0
- package/dist/learning/seed-report.js.map +1 -0
- package/dist/learning/trajectory-sources/dataclaw.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/dataclaw.js +6 -18
- package/dist/learning/trajectory-sources/dataclaw.js.map +1 -1
- package/dist/learning/trajectory-sources/entire.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/entire.js +30 -9
- package/dist/learning/trajectory-sources/entire.js.map +1 -1
- package/dist/learning/trajectory-sources/file.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/file.js.map +1 -1
- package/dist/learning/trajectory-sources/huggingface.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/huggingface.js +7 -7
- package/dist/learning/trajectory-sources/huggingface.js.map +1 -1
- package/dist/learning/trajectory-sources/in-memory.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/in-memory.js.map +1 -1
- package/dist/learning/trajectory-sources/index.d.ts +3 -4
- package/dist/learning/trajectory-sources/index.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/index.js +3 -4
- package/dist/learning/trajectory-sources/index.js.map +1 -1
- package/dist/learning/trajectory-sources/pipeline.d.ts +4 -3
- package/dist/learning/trajectory-sources/pipeline.d.ts.map +1 -1
- package/dist/learning/trajectory-sources/pipeline.js +4 -3
- package/dist/learning/trajectory-sources/pipeline.js.map +1 -1
- package/dist/learning/unified-pipeline.d.ts +486 -61
- package/dist/learning/unified-pipeline.d.ts.map +1 -1
- package/dist/learning/unified-pipeline.js +1501 -232
- package/dist/learning/unified-pipeline.js.map +1 -1
- package/dist/learning/usage-inference.d.ts.map +1 -1
- package/dist/learning/usage-inference.js +4 -9
- package/dist/learning/usage-inference.js.map +1 -1
- package/dist/learning/workspace-snapshot.d.ts +146 -0
- package/dist/learning/workspace-snapshot.d.ts.map +1 -0
- package/dist/learning/workspace-snapshot.js +269 -0
- package/dist/learning/workspace-snapshot.js.map +1 -0
- package/dist/mcp/playbook-server.d.ts +61 -2
- package/dist/mcp/playbook-server.d.ts.map +1 -1
- package/dist/mcp/playbook-server.js +179 -9
- package/dist/mcp/playbook-server.js.map +1 -1
- package/dist/memory/coherence.d.ts.map +1 -1
- package/dist/memory/coherence.js +5 -7
- package/dist/memory/coherence.js.map +1 -1
- package/dist/memory/curated-loader.d.ts +24 -0
- package/dist/memory/curated-loader.d.ts.map +1 -1
- package/dist/memory/curated-loader.js +22 -5
- package/dist/memory/curated-loader.js.map +1 -1
- package/dist/memory/evolution/llm-evolver.d.ts +36 -0
- package/dist/memory/evolution/llm-evolver.d.ts.map +1 -0
- package/dist/memory/evolution/llm-evolver.js +106 -0
- package/dist/memory/evolution/llm-evolver.js.map +1 -0
- package/dist/memory/experience.d.ts +34 -13
- package/dist/memory/experience.d.ts.map +1 -1
- package/dist/memory/experience.js +59 -48
- package/dist/memory/experience.js.map +1 -1
- package/dist/memory/extraction/conversation-extractor.d.ts +62 -0
- package/dist/memory/extraction/conversation-extractor.d.ts.map +1 -0
- package/dist/memory/extraction/conversation-extractor.js +169 -0
- package/dist/memory/extraction/conversation-extractor.js.map +1 -0
- package/dist/memory/filesystem-cascade.d.ts +116 -0
- package/dist/memory/filesystem-cascade.d.ts.map +1 -0
- package/dist/memory/filesystem-cascade.js +221 -0
- package/dist/memory/filesystem-cascade.js.map +1 -0
- package/dist/memory/index.d.ts +14 -10
- package/dist/memory/index.d.ts.map +1 -1
- package/dist/memory/index.js +15 -12
- package/dist/memory/index.js.map +1 -1
- package/dist/memory/knowledge-bank.d.ts +145 -29
- package/dist/memory/knowledge-bank.d.ts.map +1 -1
- package/dist/memory/knowledge-bank.js +511 -200
- package/dist/memory/knowledge-bank.js.map +1 -1
- package/dist/memory/knowledge-defrag.d.ts +32 -9
- package/dist/memory/knowledge-defrag.d.ts.map +1 -1
- package/dist/memory/knowledge-defrag.js +57 -49
- package/dist/memory/knowledge-defrag.js.map +1 -1
- package/dist/memory/meta-playbook-bank.d.ts +24 -0
- package/dist/memory/meta-playbook-bank.d.ts.map +1 -0
- package/dist/memory/meta-playbook-bank.js +135 -0
- package/dist/memory/meta-playbook-bank.js.map +1 -0
- package/dist/memory/meta.d.ts.map +1 -1
- package/dist/memory/meta.js +33 -14
- package/dist/memory/meta.js.map +1 -1
- package/dist/memory/playbook-applicability.d.ts +43 -0
- package/dist/memory/playbook-applicability.d.ts.map +1 -0
- package/dist/memory/playbook-applicability.js +155 -0
- package/dist/memory/playbook-applicability.js.map +1 -0
- package/dist/memory/playbook-incorporation.d.ts +24 -0
- package/dist/memory/playbook-incorporation.d.ts.map +1 -0
- package/dist/memory/playbook-incorporation.js +41 -0
- package/dist/memory/playbook-incorporation.js.map +1 -0
- package/dist/memory/playbook.d.ts +444 -7
- package/dist/memory/playbook.d.ts.map +1 -1
- package/dist/memory/playbook.js +1231 -50
- package/dist/memory/playbook.js.map +1 -1
- package/dist/memory/procedural-skill.d.ts +142 -0
- package/dist/memory/procedural-skill.d.ts.map +1 -0
- package/dist/memory/procedural-skill.js +846 -0
- package/dist/memory/procedural-skill.js.map +1 -0
- package/dist/memory/search-provider.d.ts +8 -3
- package/dist/memory/search-provider.d.ts.map +1 -1
- package/dist/memory/search-providers/embedding.d.ts +64 -0
- package/dist/memory/search-providers/embedding.d.ts.map +1 -0
- package/dist/memory/search-providers/embedding.js +96 -0
- package/dist/memory/search-providers/embedding.js.map +1 -0
- package/dist/memory/search-providers/hybrid.d.ts +74 -0
- package/dist/memory/search-providers/hybrid.d.ts.map +1 -0
- package/dist/memory/search-providers/hybrid.js +94 -0
- package/dist/memory/search-providers/hybrid.js.map +1 -0
- package/dist/memory/search-providers/index.d.ts +3 -0
- package/dist/memory/search-providers/index.d.ts.map +1 -1
- package/dist/memory/search-providers/index.js +3 -0
- package/dist/memory/search-providers/index.js.map +1 -1
- package/dist/memory/search-providers/keyword-expanding.d.ts +22 -0
- package/dist/memory/search-providers/keyword-expanding.d.ts.map +1 -0
- package/dist/memory/search-providers/keyword-expanding.js +35 -0
- package/dist/memory/search-providers/keyword-expanding.js.map +1 -0
- package/dist/memory/search-providers/minimem.d.ts +16 -4
- package/dist/memory/search-providers/minimem.d.ts.map +1 -1
- package/dist/memory/search-providers/minimem.js +14 -3
- package/dist/memory/search-providers/minimem.js.map +1 -1
- package/dist/memory/search-providers/text-similarity.d.ts +2 -4
- package/dist/memory/search-providers/text-similarity.d.ts.map +1 -1
- package/dist/memory/search-providers/text-similarity.js.map +1 -1
- package/dist/memory/seeds/meta-playbook-seeds.d.ts +27 -0
- package/dist/memory/seeds/meta-playbook-seeds.d.ts.map +1 -0
- package/dist/memory/seeds/meta-playbook-seeds.js +560 -0
- package/dist/memory/seeds/meta-playbook-seeds.js.map +1 -0
- package/dist/memory/seeds/meta-strategy-seeds.d.ts +23 -0
- package/dist/memory/seeds/meta-strategy-seeds.d.ts.map +1 -0
- package/dist/memory/seeds/meta-strategy-seeds.js +82 -0
- package/dist/memory/seeds/meta-strategy-seeds.js.map +1 -0
- package/dist/memory/seeds/seed-loader.d.ts +34 -0
- package/dist/memory/seeds/seed-loader.d.ts.map +1 -0
- package/dist/memory/seeds/seed-loader.js +60 -0
- package/dist/memory/seeds/seed-loader.js.map +1 -0
- package/dist/memory/skill-exporter.d.ts.map +1 -1
- package/dist/memory/skill-exporter.js +1 -2
- package/dist/memory/skill-exporter.js.map +1 -1
- package/dist/memory/source-resolver.d.ts.map +1 -1
- package/dist/memory/source-resolver.js +7 -8
- package/dist/memory/source-resolver.js.map +1 -1
- package/dist/memory/system.d.ts +119 -14
- package/dist/memory/system.d.ts.map +1 -1
- package/dist/memory/system.js +416 -60
- package/dist/memory/system.js.map +1 -1
- package/dist/memory/temporal-compressor.d.ts +1 -1
- package/dist/memory/temporal-compressor.d.ts.map +1 -1
- package/dist/memory/temporal-compressor.js +26 -16
- package/dist/memory/temporal-compressor.js.map +1 -1
- package/dist/persistence/index.d.ts +2 -2
- package/dist/persistence/index.d.ts.map +1 -1
- package/dist/persistence/index.js +1 -1
- package/dist/persistence/index.js.map +1 -1
- package/dist/persistence/migrator.d.ts +34 -1
- package/dist/persistence/migrator.d.ts.map +1 -1
- package/dist/persistence/migrator.js +163 -33
- package/dist/persistence/migrator.js.map +1 -1
- package/dist/persistence/sqlite-persistence.d.ts +412 -1
- package/dist/persistence/sqlite-persistence.d.ts.map +1 -1
- package/dist/persistence/sqlite-persistence.js +1038 -47
- package/dist/persistence/sqlite-persistence.js.map +1 -1
- package/dist/runtime/backends/acp-protocol.d.ts.map +1 -1
- package/dist/runtime/backends/acp-protocol.js +2 -1
- package/dist/runtime/backends/acp-protocol.js.map +1 -1
- package/dist/runtime/backends/acp.d.ts.map +1 -1
- package/dist/runtime/backends/acp.js +5 -4
- package/dist/runtime/backends/acp.js.map +1 -1
- package/dist/runtime/backends/index.d.ts +2 -2
- package/dist/runtime/backends/index.d.ts.map +1 -1
- package/dist/runtime/backends/index.js +2 -2
- package/dist/runtime/backends/index.js.map +1 -1
- package/dist/runtime/backends/macro-agent.d.ts +17 -1
- package/dist/runtime/backends/macro-agent.d.ts.map +1 -1
- package/dist/runtime/backends/macro-agent.js +1 -1
- package/dist/runtime/backends/macro-agent.js.map +1 -1
- package/dist/runtime/backends/mock.d.ts.map +1 -1
- package/dist/runtime/backends/mock.js.map +1 -1
- package/dist/runtime/backends/subprocess.d.ts.map +1 -1
- package/dist/runtime/backends/subprocess.js +5 -4
- package/dist/runtime/backends/subprocess.js.map +1 -1
- package/dist/runtime/compute-provider.d.ts +2 -18
- package/dist/runtime/compute-provider.d.ts.map +1 -1
- package/dist/runtime/compute-provider.js +8 -8
- package/dist/runtime/compute-provider.js.map +1 -1
- package/dist/runtime/flows/learning.d.ts +2 -1
- package/dist/runtime/flows/learning.d.ts.map +1 -1
- package/dist/runtime/flows/learning.js +4 -5
- package/dist/runtime/flows/learning.js.map +1 -1
- package/dist/runtime/flows/validation.d.ts.map +1 -1
- package/dist/runtime/flows/validation.js +2 -5
- package/dist/runtime/flows/validation.js.map +1 -1
- package/dist/runtime/index.d.ts +1 -1
- package/dist/runtime/index.d.ts.map +1 -1
- package/dist/runtime/index.js.map +1 -1
- package/dist/runtime/manager.d.ts +128 -0
- package/dist/runtime/manager.d.ts.map +1 -1
- package/dist/runtime/manager.js +375 -64
- package/dist/runtime/manager.js.map +1 -1
- package/dist/runtime/types.d.ts +92 -1
- package/dist/runtime/types.d.ts.map +1 -1
- package/dist/search/evaluator.d.ts +71 -0
- package/dist/search/evaluator.d.ts.map +1 -1
- package/dist/search/evaluator.js +40 -5
- package/dist/search/evaluator.js.map +1 -1
- package/dist/search/index.d.ts +3 -5
- package/dist/search/index.d.ts.map +1 -1
- package/dist/search/index.js +4 -7
- package/dist/search/index.js.map +1 -1
- package/dist/search/moe-gate.d.ts +48 -2
- package/dist/search/moe-gate.d.ts.map +1 -1
- package/dist/search/moe-gate.js +117 -4
- package/dist/search/moe-gate.js.map +1 -1
- package/dist/search/refinement-loop.d.ts +63 -3
- package/dist/search/refinement-loop.d.ts.map +1 -1
- package/dist/search/refinement-loop.js +278 -23
- package/dist/search/refinement-loop.js.map +1 -1
- package/dist/search/refinement-types.d.ts +4 -2
- package/dist/search/refinement-types.d.ts.map +1 -1
- package/dist/search/refinement-types.js +2 -11
- package/dist/search/refinement-types.js.map +1 -1
- package/dist/search/router.d.ts +41 -34
- package/dist/search/router.d.ts.map +1 -1
- package/dist/search/router.js +156 -90
- package/dist/search/router.js.map +1 -1
- package/dist/search/solver.js +1 -1
- package/dist/search/solver.js.map +1 -1
- package/dist/search/task-difficulty.d.ts +54 -0
- package/dist/search/task-difficulty.d.ts.map +1 -0
- package/dist/search/task-difficulty.js +116 -0
- package/dist/search/task-difficulty.js.map +1 -0
- package/dist/search/tau-bench-verifier.d.ts +74 -0
- package/dist/search/tau-bench-verifier.d.ts.map +1 -0
- package/dist/search/tau-bench-verifier.js +96 -0
- package/dist/search/tau-bench-verifier.js.map +1 -0
- package/dist/search/verification-runner.d.ts.map +1 -1
- package/dist/search/verification-runner.js +6 -5
- package/dist/search/verification-runner.js.map +1 -1
- package/dist/session-bank/git-reader.d.ts.map +1 -1
- package/dist/session-bank/git-reader.js +5 -19
- package/dist/session-bank/git-reader.js.map +1 -1
- package/dist/session-bank/parser.d.ts.map +1 -1
- package/dist/session-bank/parser.js +26 -0
- package/dist/session-bank/parser.js.map +1 -1
- package/dist/session-bank/session-bank.d.ts.map +1 -1
- package/dist/session-bank/session-bank.js +2 -5
- package/dist/session-bank/session-bank.js.map +1 -1
- package/dist/session-bank/types.d.ts +24 -0
- package/dist/session-bank/types.d.ts.map +1 -1
- package/dist/surfacing/index.d.ts +3 -3
- package/dist/surfacing/index.d.ts.map +1 -1
- package/dist/surfacing/index.js +2 -2
- package/dist/surfacing/index.js.map +1 -1
- package/dist/surfacing/shared-bank.d.ts +43 -0
- package/dist/surfacing/shared-bank.d.ts.map +1 -0
- package/dist/surfacing/shared-bank.js +104 -0
- package/dist/surfacing/shared-bank.js.map +1 -0
- package/dist/surfacing/skill-library.d.ts +98 -2
- package/dist/surfacing/skill-library.d.ts.map +1 -1
- package/dist/surfacing/skill-library.js +192 -17
- package/dist/surfacing/skill-library.js.map +1 -1
- package/dist/surfacing/skill-publisher.d.ts +3 -1
- package/dist/surfacing/skill-publisher.d.ts.map +1 -1
- package/dist/surfacing/skill-publisher.js +28 -28
- package/dist/surfacing/skill-publisher.js.map +1 -1
- package/dist/surfacing/sqlite-storage-adapter.d.ts +2 -2
- package/dist/surfacing/sqlite-storage-adapter.d.ts.map +1 -1
- package/dist/surfacing/sqlite-storage-adapter.js +27 -18
- package/dist/surfacing/sqlite-storage-adapter.js.map +1 -1
- package/dist/types/analysis.d.ts +43 -0
- package/dist/types/analysis.d.ts.map +1 -0
- package/dist/types/analysis.js +10 -0
- package/dist/types/analysis.js.map +1 -0
- package/dist/types/config.d.ts +2178 -425
- package/dist/types/config.d.ts.map +1 -1
- package/dist/types/config.js +600 -88
- package/dist/types/config.js.map +1 -1
- package/dist/types/dataclaw.d.ts +3 -3
- package/dist/types/index.d.ts +4 -7
- package/dist/types/index.d.ts.map +1 -1
- package/dist/types/index.js +2 -10
- package/dist/types/index.js.map +1 -1
- package/dist/types/knowledge-graph.d.ts +1 -1
- package/dist/types/knowledge-graph.d.ts.map +1 -1
- package/dist/types/knowledge.d.ts +219 -11
- package/dist/types/knowledge.d.ts.map +1 -1
- package/dist/types/knowledge.js +90 -30
- package/dist/types/knowledge.js.map +1 -1
- package/dist/types/memory.d.ts +8 -0
- package/dist/types/memory.d.ts.map +1 -1
- package/dist/types/memory.js +12 -4
- package/dist/types/memory.js.map +1 -1
- package/dist/types/outcome.d.ts +11 -0
- package/dist/types/outcome.d.ts.map +1 -1
- package/dist/types/outcome.js +9 -0
- package/dist/types/outcome.js.map +1 -1
- package/dist/types/playbook.d.ts +63 -0
- package/dist/types/playbook.d.ts.map +1 -1
- package/dist/types/playbook.js +51 -0
- package/dist/types/playbook.js.map +1 -1
- package/dist/types/trajectory-source.d.ts +1 -1
- package/dist/types/trajectory-source.d.ts.map +1 -1
- package/dist/types/trajectory.d.ts +61 -0
- package/dist/types/trajectory.d.ts.map +1 -1
- package/dist/types/trajectory.js +25 -0
- package/dist/types/trajectory.js.map +1 -1
- package/dist/utils/clock.d.ts +36 -0
- package/dist/utils/clock.d.ts.map +1 -0
- package/dist/utils/clock.js +48 -0
- package/dist/utils/clock.js.map +1 -0
- package/dist/utils/frontmatter.d.ts.map +1 -1
- package/dist/utils/frontmatter.js +10 -2
- package/dist/utils/frontmatter.js.map +1 -1
- package/dist/utils/index.d.ts +4 -1
- package/dist/utils/index.d.ts.map +1 -1
- package/dist/utils/index.js +4 -1
- package/dist/utils/index.js.map +1 -1
- package/dist/utils/logger.d.ts +31 -0
- package/dist/utils/logger.d.ts.map +1 -0
- package/dist/utils/logger.js +50 -0
- package/dist/utils/logger.js.map +1 -0
- package/dist/utils/objectives.d.ts +45 -0
- package/dist/utils/objectives.d.ts.map +1 -0
- package/dist/utils/objectives.js +74 -0
- package/dist/utils/objectives.js.map +1 -0
- package/dist/utils/partitioned-store.d.ts.map +1 -1
- package/dist/utils/partitioned-store.js.map +1 -1
- package/dist/utils/playbook-hash.d.ts +19 -0
- package/dist/utils/playbook-hash.d.ts.map +1 -0
- package/dist/utils/playbook-hash.js +51 -0
- package/dist/utils/playbook-hash.js.map +1 -0
- package/dist/utils/similarity.d.ts.map +1 -1
- package/dist/utils/similarity.js.map +1 -1
- package/dist/utils/solver-canonicalizer.d.ts +22 -0
- package/dist/utils/solver-canonicalizer.d.ts.map +1 -0
- package/dist/utils/solver-canonicalizer.js +41 -0
- package/dist/utils/solver-canonicalizer.js.map +1 -0
- package/dist/utils/storage.d.ts.map +1 -1
- package/dist/utils/storage.js +1 -3
- package/dist/utils/storage.js.map +1 -1
- package/dist/utils/task-signature.d.ts +34 -0
- package/dist/utils/task-signature.d.ts.map +1 -0
- package/dist/utils/task-signature.js +112 -0
- package/dist/utils/task-signature.js.map +1 -0
- package/dist/utils/trajectory-helpers.d.ts.map +1 -1
- package/dist/utils/trajectory-helpers.js.map +1 -1
- package/dist/utils/validation.d.ts.map +1 -1
- package/dist/utils/validation.js.map +1 -1
- package/dist/workspace/efficacy-toolkit.d.ts.map +1 -1
- package/dist/workspace/efficacy-toolkit.js +18 -26
- package/dist/workspace/efficacy-toolkit.js.map +1 -1
- package/dist/workspace/index.d.ts +4 -3
- package/dist/workspace/index.d.ts.map +1 -1
- package/dist/workspace/index.js +5 -3
- package/dist/workspace/index.js.map +1 -1
- package/dist/workspace/runner.d.ts.map +1 -1
- package/dist/workspace/runner.js +95 -21
- package/dist/workspace/runner.js.map +1 -1
- package/dist/workspace/skill-converter.d.ts +66 -1
- package/dist/workspace/skill-converter.d.ts.map +1 -1
- package/dist/workspace/skill-converter.js +238 -19
- package/dist/workspace/skill-converter.js.map +1 -1
- package/dist/workspace/skill-import.d.ts +43 -0
- package/dist/workspace/skill-import.d.ts.map +1 -0
- package/dist/workspace/skill-import.js +40 -0
- package/dist/workspace/skill-import.js.map +1 -0
- package/dist/workspace/templates/index.d.ts +6 -8
- package/dist/workspace/templates/index.d.ts.map +1 -1
- package/dist/workspace/templates/index.js +7 -11
- package/dist/workspace/templates/index.js.map +1 -1
- package/dist/workspace/templates/knowledge-defrag.d.ts.map +1 -1
- package/dist/workspace/templates/knowledge-defrag.js +0 -2
- package/dist/workspace/templates/knowledge-defrag.js.map +1 -1
- package/dist/workspace/templates/knowledge-extraction.d.ts +11 -0
- package/dist/workspace/templates/knowledge-extraction.d.ts.map +1 -1
- package/dist/workspace/templates/knowledge-extraction.js +11 -6
- package/dist/workspace/templates/knowledge-extraction.js.map +1 -1
- package/dist/workspace/templates/playbook-conditioning.d.ts +38 -0
- package/dist/workspace/templates/playbook-conditioning.d.ts.map +1 -0
- package/dist/workspace/templates/playbook-conditioning.js +147 -0
- package/dist/workspace/templates/playbook-conditioning.js.map +1 -0
- package/dist/workspace/templates/playbook-decay-detection.d.ts.map +1 -1
- package/dist/workspace/templates/playbook-decay-detection.js +24 -18
- package/dist/workspace/templates/playbook-decay-detection.js.map +1 -1
- package/dist/workspace/templates/playbook-efficacy-audit.d.ts.map +1 -1
- package/dist/workspace/templates/playbook-efficacy-audit.js +26 -11
- package/dist/workspace/templates/playbook-efficacy-audit.js.map +1 -1
- package/dist/workspace/templates/playbook-extraction.d.ts.map +1 -1
- package/dist/workspace/templates/playbook-extraction.js +75 -15
- package/dist/workspace/templates/playbook-extraction.js.map +1 -1
- package/dist/workspace/templates/playbook-lifecycle-review.d.ts.map +1 -1
- package/dist/workspace/templates/playbook-lifecycle-review.js +12 -8
- package/dist/workspace/templates/playbook-lifecycle-review.js.map +1 -1
- package/dist/workspace/templates/refinement-analysis.d.ts.map +1 -1
- package/dist/workspace/templates/refinement-analysis.js +11 -5
- package/dist/workspace/templates/refinement-analysis.js.map +1 -1
- package/dist/workspace/templates/skill-enrichment.d.ts.map +1 -1
- package/dist/workspace/templates/skill-enrichment.js +7 -11
- package/dist/workspace/templates/skill-enrichment.js.map +1 -1
- package/dist/workspace/templates/solution-evaluation.d.ts.map +1 -1
- package/dist/workspace/templates/solution-evaluation.js +15 -8
- package/dist/workspace/templates/solution-evaluation.js.map +1 -1
- package/dist/workspace/templates/trajectory-analysis.d.ts.map +1 -1
- package/dist/workspace/templates/trajectory-analysis.js +1 -3
- package/dist/workspace/templates/trajectory-analysis.js.map +1 -1
- package/dist/workspace/templates/usage-inference.d.ts.map +1 -1
- package/dist/workspace/templates/usage-inference.js +9 -7
- package/dist/workspace/templates/usage-inference.js.map +1 -1
- package/dist/workspace/types.d.ts +32 -2
- package/dist/workspace/types.d.ts.map +1 -1
- package/dist/workspace/types.js +29 -0
- package/dist/workspace/types.js.map +1 -1
- package/package.json +43 -19
- package/playbooks/agent-ops/long-task-checkpointing.json +54 -0
- package/playbooks/agent-ops/progressive-context-gathering.json +56 -0
- package/playbooks/agent-ops/safe-destructive-operations.json +59 -0
- package/playbooks/agent-ops/scope-discipline.json +56 -0
- package/playbooks/agent-ops/stall-detection-strategy-switch.json +55 -0
- package/playbooks/agent-ops/verification-before-completion.json +54 -0
- package/playbooks/alfworld/clean-and-place.json +35 -0
- package/playbooks/alfworld/cool-and-place.json +35 -0
- package/playbooks/alfworld/examine-in-light.json +32 -0
- package/playbooks/alfworld/heat-and-place.json +35 -0
- package/playbooks/alfworld/pick-and-place.json +37 -0
- package/playbooks/alfworld/pick-two-and-place.json +35 -0
- package/.claude/settings.json +0 -197
- package/.entire/settings.json +0 -4
- package/.sessionlog/settings.json +0 -4
- package/CLAUDE.md +0 -233
- package/SKILL.md +0 -193
- package/docs/DESIGN-agentic-workspace.md +0 -2057
- package/docs/DESIGN-semantic-memory-knowledge-bank.md +0 -1789
- package/docs/DESIGN-session-bank.md +0 -1134
- package/docs/DESIGN-team-extraction-pipeline.md +0 -1369
- package/docs/PROPOSAL-team-aware-learning.md +0 -1080
- package/docs/RESEARCH-semantic-memory-knowledge-structures.md +0 -517
- package/docs/RESEARCH-team-trajectory-learning.md +0 -553
- package/gaps.md +0 -204
- package/scripts/migrate-to-playbooks.ts +0 -307
- package/src/agents/index.ts +0 -14
- package/src/agents/mock-provider.ts +0 -93
- package/src/agents/types.ts +0 -137
- package/src/atlas.ts +0 -1198
- package/src/bin/cli-utils.ts +0 -220
- package/src/bin/cognitive-core.ts +0 -162
- package/src/bin/commands/kb.ts +0 -266
- package/src/bin/commands/learn.ts +0 -100
- package/src/bin/commands/legacy.ts +0 -182
- package/src/bin/commands/run.ts +0 -113
- package/src/bin/commands/sessions.ts +0 -221
- package/src/bin/commands/skills.ts +0 -146
- package/src/embeddings/bm25.ts +0 -337
- package/src/embeddings/index.ts +0 -42
- package/src/embeddings/inverted-index.ts +0 -134
- package/src/embeddings/manager.ts +0 -293
- package/src/embeddings/provider.ts +0 -318
- package/src/embeddings/vector-store.ts +0 -365
- package/src/factory.ts +0 -280
- package/src/index.ts +0 -495
- package/src/learning/analyzer.ts +0 -319
- package/src/learning/effectiveness.ts +0 -428
- package/src/learning/energy-evaluator.ts +0 -282
- package/src/learning/healing-orchestrator.ts +0 -383
- package/src/learning/index.ts +0 -150
- package/src/learning/instant-loop.ts +0 -357
- package/src/learning/knowledge-extractor.ts +0 -470
- package/src/learning/maintenance-scheduler.ts +0 -271
- package/src/learning/meta-learner.ts +0 -552
- package/src/learning/playbook-extractor.ts +0 -702
- package/src/learning/reflexion-generator.ts +0 -273
- package/src/learning/team-ingester.ts +0 -499
- package/src/learning/team-meta-learner.ts +0 -558
- package/src/learning/team-pipeline.ts +0 -364
- package/src/learning/trajectory-sources/dataclaw.ts +0 -403
- package/src/learning/trajectory-sources/entire.ts +0 -251
- package/src/learning/trajectory-sources/file.ts +0 -136
- package/src/learning/trajectory-sources/huggingface.ts +0 -248
- package/src/learning/trajectory-sources/in-memory.ts +0 -61
- package/src/learning/trajectory-sources/index.ts +0 -15
- package/src/learning/trajectory-sources/pipeline.ts +0 -69
- package/src/learning/unified-pipeline.ts +0 -1191
- package/src/learning/usage-inference.ts +0 -372
- package/src/mcp/index.ts +0 -12
- package/src/mcp/playbook-server.ts +0 -565
- package/src/memory/candidate-retrieval.ts +0 -72
- package/src/memory/causal-store.ts +0 -273
- package/src/memory/coherence.ts +0 -252
- package/src/memory/curated-loader.ts +0 -213
- package/src/memory/experience.ts +0 -682
- package/src/memory/graph-layers/base.ts +0 -184
- package/src/memory/graph-layers/causal.ts +0 -16
- package/src/memory/graph-layers/entity.ts +0 -16
- package/src/memory/graph-layers/index.ts +0 -5
- package/src/memory/graph-layers/semantic.ts +0 -16
- package/src/memory/graph-layers/temporal.ts +0 -16
- package/src/memory/index.ts +0 -131
- package/src/memory/knowledge-bank.ts +0 -1317
- package/src/memory/knowledge-defrag.ts +0 -329
- package/src/memory/knowledge-graph.ts +0 -361
- package/src/memory/meta.ts +0 -488
- package/src/memory/playbook.ts +0 -676
- package/src/memory/reasoning-bank.ts +0 -458
- package/src/memory/reflexion.ts +0 -122
- package/src/memory/search-provider.ts +0 -35
- package/src/memory/search-providers/index.ts +0 -3
- package/src/memory/search-providers/minimem.ts +0 -84
- package/src/memory/search-providers/text-similarity.ts +0 -35
- package/src/memory/skill-exporter.ts +0 -357
- package/src/memory/source-resolver.ts +0 -422
- package/src/memory/system.ts +0 -620
- package/src/memory/team-experience.ts +0 -604
- package/src/memory/temporal-compressor.ts +0 -409
- package/src/persistence/index.ts +0 -37
- package/src/persistence/migrator.ts +0 -298
- package/src/persistence/serializers.ts +0 -79
- package/src/persistence/sqlite-persistence.ts +0 -925
- package/src/runtime/backends/acp-protocol.ts +0 -231
- package/src/runtime/backends/acp.ts +0 -239
- package/src/runtime/backends/index.ts +0 -38
- package/src/runtime/backends/macro-agent.ts +0 -177
- package/src/runtime/backends/mock.ts +0 -218
- package/src/runtime/backends/subprocess.ts +0 -356
- package/src/runtime/compute-provider.ts +0 -206
- package/src/runtime/flows/learning.ts +0 -166
- package/src/runtime/flows/validation.ts +0 -381
- package/src/runtime/index.ts +0 -75
- package/src/runtime/manager.ts +0 -685
- package/src/runtime/types.ts +0 -198
- package/src/search/evaluator.ts +0 -499
- package/src/search/index.ts +0 -78
- package/src/search/moe-gate.ts +0 -304
- package/src/search/refinement-loop.ts +0 -457
- package/src/search/refinement-types.ts +0 -159
- package/src/search/router.ts +0 -368
- package/src/search/solver.ts +0 -303
- package/src/search/team-router.ts +0 -459
- package/src/search/verification-runner.ts +0 -570
- package/src/session-bank/git-reader.ts +0 -200
- package/src/session-bank/index.ts +0 -26
- package/src/session-bank/parser.ts +0 -548
- package/src/session-bank/session-bank.ts +0 -485
- package/src/session-bank/types.ts +0 -176
- package/src/surfacing/index.ts +0 -29
- package/src/surfacing/publisher.ts +0 -23
- package/src/surfacing/skill-library.ts +0 -625
- package/src/surfacing/skill-publisher.ts +0 -257
- package/src/surfacing/sqlite-storage-adapter.ts +0 -285
- package/src/surfacing/team-skill-library.ts +0 -610
- package/src/types/config.ts +0 -661
- package/src/types/dataclaw.ts +0 -99
- package/src/types/huggingface-transformers.d.ts +0 -12
- package/src/types/index.ts +0 -293
- package/src/types/knowledge-graph.ts +0 -246
- package/src/types/knowledge.ts +0 -388
- package/src/types/memory.ts +0 -300
- package/src/types/meta.ts +0 -218
- package/src/types/outcome.ts +0 -66
- package/src/types/playbook.ts +0 -215
- package/src/types/step.ts +0 -40
- package/src/types/task.ts +0 -52
- package/src/types/team-meta.ts +0 -212
- package/src/types/team-playbook.ts +0 -384
- package/src/types/team-trajectory.ts +0 -673
- package/src/types/trajectory-source.ts +0 -47
- package/src/types/trajectory.ts +0 -80
- package/src/utils/error-classifier.ts +0 -113
- package/src/utils/frontmatter.ts +0 -118
- package/src/utils/index.ts +0 -63
- package/src/utils/partitioned-store.ts +0 -299
- package/src/utils/similarity.ts +0 -139
- package/src/utils/storage.ts +0 -249
- package/src/utils/trajectory-helpers.ts +0 -79
- package/src/utils/validation.ts +0 -286
- package/src/workspace/efficacy-toolkit.ts +0 -496
- package/src/workspace/index.ts +0 -77
- package/src/workspace/runner.ts +0 -281
- package/src/workspace/skill-converter.ts +0 -288
- package/src/workspace/templates/index.ts +0 -95
- package/src/workspace/templates/knowledge-defrag.ts +0 -223
- package/src/workspace/templates/knowledge-extraction.ts +0 -337
- package/src/workspace/templates/meta-reflection.ts +0 -208
- package/src/workspace/templates/playbook-decay-detection.ts +0 -272
- package/src/workspace/templates/playbook-efficacy-audit.ts +0 -246
- package/src/workspace/templates/playbook-extraction.ts +0 -240
- package/src/workspace/templates/playbook-lifecycle-review.ts +0 -274
- package/src/workspace/templates/refinement-analysis.ts +0 -162
- package/src/workspace/templates/skill-enrichment.ts +0 -275
- package/src/workspace/templates/solution-evaluation.ts +0 -199
- package/src/workspace/templates/team-playbook-extraction.ts +0 -631
- package/src/workspace/templates/team-trajectory-analysis.ts +0 -564
- package/src/workspace/templates/trajectory-analysis.ts +0 -234
- package/src/workspace/templates/usage-inference.ts +0 -163
- package/src/workspace/types.ts +0 -185
- package/tests/atlas-knowledge.test.ts +0 -165
- package/tests/embeddings/bm25.test.ts +0 -130
- package/tests/embeddings/inverted-index.test.ts +0 -138
- package/tests/embeddings/manager.test.ts +0 -205
- package/tests/feature-toggles.test.ts +0 -275
- package/tests/fixtures/behavioral-trajectories.ts +0 -210
- package/tests/gap-fixes.test.ts +0 -1116
- package/tests/integration/atlas.test.ts +0 -266
- package/tests/integration/cli-e2e.test.ts +0 -621
- package/tests/integration/curated-sources-e2e.test.ts +0 -502
- package/tests/integration/dataclaw-e2e.test.ts +0 -559
- package/tests/integration/e2e.test.ts +0 -1337
- package/tests/integration/entire-e2e.test.ts +0 -376
- package/tests/integration/huggingface-e2e.test.ts +0 -627
- package/tests/integration/persistence-e2e.test.ts +0 -741
- package/tests/integration/phase-e2e.test.ts +0 -1143
- package/tests/integration/pipeline-data-correctness.test.ts +0 -794
- package/tests/integration/session-bank.test.ts +0 -231
- package/tests/integration/sessionlog-e2e.test.ts +0 -329
- package/tests/integration/skill-tree-wiring.test.ts +0 -152
- package/tests/integration/unified-pipeline-e2e.test.ts +0 -634
- package/tests/learning/analyzer.test.ts +0 -426
- package/tests/learning/dataclaw-trajectory-source.test.ts +0 -642
- package/tests/learning/effectiveness.test.ts +0 -542
- package/tests/learning/energy-evaluator.test.ts +0 -180
- package/tests/learning/entire-trajectory-source.test.ts +0 -380
- package/tests/learning/healing-orchestrator.test.ts +0 -269
- package/tests/learning/huggingface-trajectory-source.test.ts +0 -817
- package/tests/learning/instant-loop.test.ts +0 -243
- package/tests/learning/knowledge-extractor.test.ts +0 -491
- package/tests/learning/maintenance-scheduler.test.ts +0 -191
- package/tests/learning/meta-learner.test.ts +0 -418
- package/tests/learning/pipeline-memory-updates.test.ts +0 -721
- package/tests/learning/playbook-extractor-provenance.test.ts +0 -114
- package/tests/learning/reflexion-generator.test.ts +0 -411
- package/tests/learning/team-ingester.test.ts +0 -349
- package/tests/learning/team-meta-learner.test.ts +0 -618
- package/tests/learning/team-pipeline.test.ts +0 -334
- package/tests/learning/trajectory-sources.test.ts +0 -320
- package/tests/learning/unified-pipeline-efficacy.test.ts +0 -232
- package/tests/learning/unified-pipeline.test.ts +0 -322
- package/tests/learning/usage-inference.test.ts +0 -254
- package/tests/mcp/playbook-server.test.ts +0 -257
- package/tests/memory/candidate-retrieval.test.ts +0 -167
- package/tests/memory/causal-store.test.ts +0 -276
- package/tests/memory/coherence.test.ts +0 -232
- package/tests/memory/compound-engineering-seed.test.ts +0 -338
- package/tests/memory/curated-loader-extended.test.ts +0 -225
- package/tests/memory/experience.test.ts +0 -203
- package/tests/memory/graph-layers/semantic.test.ts +0 -219
- package/tests/memory/knowledge-bank-extraction.test.ts +0 -558
- package/tests/memory/knowledge-bank.test.ts +0 -705
- package/tests/memory/knowledge-defrag.test.ts +0 -366
- package/tests/memory/knowledge-evolution.test.ts +0 -563
- package/tests/memory/knowledge-graph.test.ts +0 -492
- package/tests/memory/knowledge-inbox.test.ts +0 -258
- package/tests/memory/knowledge-minimem.test.ts +0 -251
- package/tests/memory/meta.test.ts +0 -399
- package/tests/memory/playbook-quality-validation.test.ts +0 -430
- package/tests/memory/playbook.test.ts +0 -644
- package/tests/memory/provenance.test.ts +0 -648
- package/tests/memory/reasoning-bank.test.ts +0 -239
- package/tests/memory/reflexion.test.ts +0 -166
- package/tests/memory/skill-exporter.test.ts +0 -475
- package/tests/memory/source-resolver.test.ts +0 -700
- package/tests/memory/system.test.ts +0 -330
- package/tests/memory/team-experience.test.ts +0 -411
- package/tests/memory/temporal-compressor.test.ts +0 -318
- package/tests/persistence/migrator.test.ts +0 -1009
- package/tests/persistence/sqlite-persistence.test.ts +0 -635
- package/tests/runtime/agent-manager.test.ts +0 -517
- package/tests/runtime/compute-provider.test.ts +0 -288
- package/tests/runtime/delegate.test.ts +0 -354
- package/tests/runtime/macro-agent-backend.test.ts +0 -266
- package/tests/runtime/mock-backend.test.ts +0 -248
- package/tests/search/evaluator.test.ts +0 -257
- package/tests/search/moe-gate.test.ts +0 -250
- package/tests/search/refinement-loop.test.ts +0 -477
- package/tests/search/refinement.test.ts +0 -267
- package/tests/search/router.test.ts +0 -506
- package/tests/search/team-router.test.ts +0 -376
- package/tests/search/verification-runner.test.ts +0 -357
- package/tests/session-bank/fixtures/multi-tool-session/full.jsonl +0 -12
- package/tests/session-bank/fixtures/multi-tool-session/metadata.json +0 -28
- package/tests/session-bank/fixtures/root-metadata.json +0 -18
- package/tests/session-bank/fixtures/sessionlog-root-metadata.json +0 -16
- package/tests/session-bank/fixtures/sessionlog-session/full.jsonl +0 -6
- package/tests/session-bank/fixtures/sessionlog-session/metadata.json +0 -55
- package/tests/session-bank/fixtures/simple-session/full.jsonl +0 -6
- package/tests/session-bank/fixtures/simple-session/metadata.json +0 -38
- package/tests/session-bank/git-reader.test.ts +0 -232
- package/tests/session-bank/parser.test.ts +0 -453
- package/tests/session-bank/session-bank.test.ts +0 -546
- package/tests/surfacing/skill-library.test.ts +0 -297
- package/tests/surfacing/skill-publisher.test.ts +0 -437
- package/tests/surfacing/sqlite-storage-adapter.test.ts +0 -206
- package/tests/surfacing/team-skill-library.test.ts +0 -444
- package/tests/types/outcome.test.ts +0 -147
- package/tests/types/step.test.ts +0 -133
- package/tests/types/task.test.ts +0 -158
- package/tests/types/team-meta.test.ts +0 -147
- package/tests/types/team-playbook.test.ts +0 -246
- package/tests/types/team-trajectory.test.ts +0 -557
- package/tests/types/trajectory.test.ts +0 -253
- package/tests/utils/error-classifier.test.ts +0 -149
- package/tests/utils/frontmatter.test.ts +0 -208
- package/tests/utils/partitioned-store.test.ts +0 -230
- package/tests/utils/similarity.test.ts +0 -188
- package/tests/utils/trajectory-helpers.test.ts +0 -163
- package/tests/utils/validation.test.ts +0 -252
- package/tests/workspace/efficacy-toolkit.test.ts +0 -404
- package/tests/workspace/full-flow.test.ts +0 -845
- package/tests/workspace/manager.test.ts +0 -215
- package/tests/workspace/runner.test.ts +0 -336
- package/tests/workspace/skill-converter.test.ts +0 -205
- package/tests/workspace/templates/knowledge-extraction.test.ts +0 -235
- package/tests/workspace/templates/playbook-efficacy.test.ts +0 -377
- package/tests/workspace/templates/team-playbook-extraction.test.ts +0 -341
- package/tests/workspace/templates/team-trajectory-analysis.test.ts +0 -417
- package/tsconfig.json +0 -25
- package/vitest.config.ts +0 -22
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
import { type Task } from '../../types/index.js';
|
|
2
|
+
import type { EvalTask } from '../types.js';
|
|
3
|
+
import type { ExtendedLiveBenchmark, BenchmarkVerdict } from './live-benchmark.js';
|
|
4
|
+
import type { BenchmarkDelegate } from './live-runner.js';
|
|
5
|
+
import { EnvBridge } from './env-bridge.js';
|
|
6
|
+
export declare class SkillLearnBenchBenchmark implements ExtendedLiveBenchmark {
|
|
7
|
+
readonly name = "skilllearnbench";
|
|
8
|
+
readonly interactive = false;
|
|
9
|
+
readonly preserveOrder = true;
|
|
10
|
+
readonly playbookMinTrajectories = 3;
|
|
11
|
+
private bridge;
|
|
12
|
+
private readonly iterations;
|
|
13
|
+
constructor(opts?: {
|
|
14
|
+
iterations?: number;
|
|
15
|
+
});
|
|
16
|
+
setup(): Promise<void>;
|
|
17
|
+
teardown(): Promise<void>;
|
|
18
|
+
private getBridge;
|
|
19
|
+
/**
|
|
20
|
+
* Load tasks as (task, iteration) pairs. Each base task is repeated
|
|
21
|
+
* `iterations` times. Ordering: all iterations for task 0, then task 1, etc.
|
|
22
|
+
* Base tasks are shuffled by seed before flattening so different seeds produce
|
|
23
|
+
* different orderings (reproducible per seed).
|
|
24
|
+
*/
|
|
25
|
+
loadTasks(opts: {
|
|
26
|
+
count: number;
|
|
27
|
+
seed: number;
|
|
28
|
+
}): Promise<EvalTask[]>;
|
|
29
|
+
/**
|
|
30
|
+
* Verify via bridge — calls the three-level evaluation (skill quality,
|
|
31
|
+
* trajectory quality, task outcome). Returns correct if overall_score > 0.5.
|
|
32
|
+
*/
|
|
33
|
+
verify(_task: Task, _output: string): BenchmarkVerdict;
|
|
34
|
+
getBridgeInstance(): EnvBridge | null;
|
|
35
|
+
delegateFactory: (_benchmark: ExtendedLiveBenchmark, opts?: {
|
|
36
|
+
model?: string;
|
|
37
|
+
timeoutMs?: number;
|
|
38
|
+
}) => BenchmarkDelegate;
|
|
39
|
+
}
|
|
40
|
+
export declare function createSkillLearnBenchBenchmark(opts?: {
|
|
41
|
+
iterations?: number;
|
|
42
|
+
}): SkillLearnBenchBenchmark;
|
|
43
|
+
//# sourceMappingURL=skilllearnbench.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skilllearnbench.d.ts","sourceRoot":"","sources":["../../../src/eval/live/skilllearnbench.ts"],"names":[],"mappings":"AAEA,OAAO,EAAc,KAAK,IAAI,EAAE,MAAM,sBAAsB,CAAC;AAC7D,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAC;AAC5C,OAAO,KAAK,EAAE,qBAAqB,EAAE,gBAAgB,EAAE,MAAM,qBAAqB,CAAC;AACnF,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,kBAAkB,CAAC;AAC1D,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAoC5C,qBAAa,wBAAyB,YAAW,qBAAqB;IACpE,QAAQ,CAAC,IAAI,qBAAqB;IAClC,QAAQ,CAAC,WAAW,SAAS;IAC7B,QAAQ,CAAC,aAAa,QAAQ;IAC9B,QAAQ,CAAC,uBAAuB,KAAK;IAErC,OAAO,CAAC,MAAM,CAA0B;IACxC,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAS;gBAExB,IAAI,GAAE;QAAE,UAAU,CAAC,EAAE,MAAM,CAAA;KAAO;IAIxC,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAItB,QAAQ,IAAI,OAAO,CAAC,IAAI,CAAC;YAKjB,SAAS;IAiBvB;;;;;OAKG;IACG,SAAS,CAAC,IAAI,EAAE;QAAE,KAAK,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,GAAG,OAAO,CAAC,QAAQ,EAAE,CAAC;IA8C3E;;;OAGG;IACH,MAAM,CAAC,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,GAAG,gBAAgB;IAItD,iBAAiB,IAAI,SAAS,GAAG,IAAI;IAIrC,eAAe,GACb,YAAY,qBAAqB,EACjC,OAAO;QAAE,KAAK,CAAC,EAAE,MAAM,CAAC;QAAC,SAAS,CAAC,EAAE,MAAM,CAAA;KAAE,KAC5C,iBAAiB,CAElB;CACH;AAED,wBAAgB,8BAA8B,CAAC,IAAI,CAAC,EAAE;IACpD,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB,GAAG,wBAAwB,CAE3B"}
|
|
@@ -0,0 +1,154 @@
|
|
|
1
|
+
import { join, dirname } from 'node:path';
|
|
2
|
+
import { fileURLToPath } from 'node:url';
|
|
3
|
+
import { createTask } from '../../types/index.js';
|
|
4
|
+
import { EnvBridge } from './env-bridge.js';
|
|
5
|
+
import { ClaudeBenchmarkDelegate } from './claude-delegate.js';
|
|
6
|
+
import { seededRng } from '../rng.js';
|
|
7
|
+
export class SkillLearnBenchBenchmark {
|
|
8
|
+
name = 'skilllearnbench';
|
|
9
|
+
interactive = false;
|
|
10
|
+
preserveOrder = true;
|
|
11
|
+
playbookMinTrajectories = 3; // iterative skill generation — extract & reuse playbooks
|
|
12
|
+
bridge = null;
|
|
13
|
+
iterations;
|
|
14
|
+
constructor(opts = {}) {
|
|
15
|
+
this.iterations = opts.iterations ?? 5;
|
|
16
|
+
}
|
|
17
|
+
async setup() {
|
|
18
|
+
await this.getBridge();
|
|
19
|
+
}
|
|
20
|
+
async teardown() {
|
|
21
|
+
await this.bridge?.close();
|
|
22
|
+
this.bridge = null;
|
|
23
|
+
}
|
|
24
|
+
async getBridge() {
|
|
25
|
+
if (this.bridge)
|
|
26
|
+
return this.bridge;
|
|
27
|
+
const { command, args } = resolveBridgeCommand();
|
|
28
|
+
const bridge = new EnvBridge(command, args);
|
|
29
|
+
try {
|
|
30
|
+
await bridge.start();
|
|
31
|
+
}
|
|
32
|
+
catch (err) {
|
|
33
|
+
throw new Error(`SkillLearnBench bridge failed to start (${err.message}).\n` +
|
|
34
|
+
`Install the SkillLearnBench Python package and set SKILLLEARNBENCH_BRIDGE\n` +
|
|
35
|
+
`or SKILLLEARNBENCH_REPO. See docs/DESIGN-evaluation-benchmarks.md.`);
|
|
36
|
+
}
|
|
37
|
+
this.bridge = bridge;
|
|
38
|
+
return bridge;
|
|
39
|
+
}
|
|
40
|
+
/**
|
|
41
|
+
* Load tasks as (task, iteration) pairs. Each base task is repeated
|
|
42
|
+
* `iterations` times. Ordering: all iterations for task 0, then task 1, etc.
|
|
43
|
+
* Base tasks are shuffled by seed before flattening so different seeds produce
|
|
44
|
+
* different orderings (reproducible per seed).
|
|
45
|
+
*/
|
|
46
|
+
async loadTasks(opts) {
|
|
47
|
+
const bridge = await this.getBridge();
|
|
48
|
+
const response = await bridge.send({ cmd: 'list' });
|
|
49
|
+
// Seeded shuffle of base tasks so different seeds produce different orderings.
|
|
50
|
+
const baseTasks = [...response.tasks];
|
|
51
|
+
const rng = seededRng('skilllearnbench-loadtasks', opts.seed);
|
|
52
|
+
for (let i = baseTasks.length - 1; i > 0; i--) {
|
|
53
|
+
const j = Math.floor(rng() * (i + 1));
|
|
54
|
+
[baseTasks[i], baseTasks[j]] = [baseTasks[j], baseTasks[i]];
|
|
55
|
+
}
|
|
56
|
+
const evalTasks = [];
|
|
57
|
+
let index = 0;
|
|
58
|
+
for (const baseTask of baseTasks) {
|
|
59
|
+
for (let iter = 0; iter < this.iterations; iter++) {
|
|
60
|
+
evalTasks.push({
|
|
61
|
+
task: createTask({
|
|
62
|
+
id: `skilllearn-${baseTask.id}-iter${iter}`,
|
|
63
|
+
domain: baseTask.domain,
|
|
64
|
+
description: baseTask.description,
|
|
65
|
+
metadata: {
|
|
66
|
+
skillTag: baseTask.domain,
|
|
67
|
+
taskId: baseTask.id,
|
|
68
|
+
iteration: iter,
|
|
69
|
+
totalIterations: this.iterations,
|
|
70
|
+
evalIndex: index,
|
|
71
|
+
},
|
|
72
|
+
}),
|
|
73
|
+
skillTag: baseTask.domain,
|
|
74
|
+
index,
|
|
75
|
+
coldProb: 0,
|
|
76
|
+
warmProb: 0,
|
|
77
|
+
family: baseTask.id,
|
|
78
|
+
iteration: iter,
|
|
79
|
+
});
|
|
80
|
+
index++;
|
|
81
|
+
if (index >= opts.count)
|
|
82
|
+
break;
|
|
83
|
+
}
|
|
84
|
+
if (index >= opts.count)
|
|
85
|
+
break;
|
|
86
|
+
}
|
|
87
|
+
return evalTasks;
|
|
88
|
+
}
|
|
89
|
+
/**
|
|
90
|
+
* Verify via bridge — calls the three-level evaluation (skill quality,
|
|
91
|
+
* trajectory quality, task outcome). Returns correct if overall_score > 0.5.
|
|
92
|
+
*/
|
|
93
|
+
verify(_task, _output) {
|
|
94
|
+
return { correct: false };
|
|
95
|
+
}
|
|
96
|
+
getBridgeInstance() {
|
|
97
|
+
return this.bridge;
|
|
98
|
+
}
|
|
99
|
+
delegateFactory = (_benchmark, opts) => {
|
|
100
|
+
return new SkillLearnDelegate(this, opts);
|
|
101
|
+
};
|
|
102
|
+
}
|
|
103
|
+
export function createSkillLearnBenchBenchmark(opts) {
|
|
104
|
+
return new SkillLearnBenchBenchmark(opts);
|
|
105
|
+
}
|
|
106
|
+
/**
|
|
107
|
+
* Single-turn delegate with bridge-side verification.
|
|
108
|
+
* After each solve, calls the bridge to get the three-level quality scores.
|
|
109
|
+
*/
|
|
110
|
+
class SkillLearnDelegate extends ClaudeBenchmarkDelegate {
|
|
111
|
+
bench;
|
|
112
|
+
constructor(bench, opts) {
|
|
113
|
+
super(bench, opts);
|
|
114
|
+
this.bench = bench;
|
|
115
|
+
}
|
|
116
|
+
async execute(prompt, options) {
|
|
117
|
+
const result = await super.execute(prompt, options);
|
|
118
|
+
const bridge = this.bench.getBridgeInstance();
|
|
119
|
+
if (bridge) {
|
|
120
|
+
try {
|
|
121
|
+
const verification = await bridge.send({
|
|
122
|
+
cmd: 'verify',
|
|
123
|
+
output: result.output,
|
|
124
|
+
});
|
|
125
|
+
return {
|
|
126
|
+
...result,
|
|
127
|
+
success: verification.task_outcome,
|
|
128
|
+
structured: {
|
|
129
|
+
...(result.structured ?? {}),
|
|
130
|
+
skillQuality: verification.skill_quality,
|
|
131
|
+
trajectoryQuality: verification.trajectory_quality,
|
|
132
|
+
overallScore: verification.overall_score,
|
|
133
|
+
},
|
|
134
|
+
};
|
|
135
|
+
}
|
|
136
|
+
catch {
|
|
137
|
+
// Bridge verification failed; fall back to base result
|
|
138
|
+
}
|
|
139
|
+
}
|
|
140
|
+
return result;
|
|
141
|
+
}
|
|
142
|
+
}
|
|
143
|
+
function resolveBridgeCommand() {
|
|
144
|
+
const env = process.env.SKILLLEARNBENCH_BRIDGE;
|
|
145
|
+
if (env && env.trim()) {
|
|
146
|
+
const parts = env.trim().split(/\s+/);
|
|
147
|
+
return { command: parts[0], args: parts.slice(1) };
|
|
148
|
+
}
|
|
149
|
+
const here = dirname(fileURLToPath(import.meta.url));
|
|
150
|
+
const srcDir = here.includes('/dist/') ? here.replace('/dist/', '/src/') : here;
|
|
151
|
+
const script = join(srcDir, 'bridges', 'skilllearnbench_bridge.py');
|
|
152
|
+
return { command: process.env.PYTHON ?? 'python3', args: [script] };
|
|
153
|
+
}
|
|
154
|
+
//# sourceMappingURL=skilllearnbench.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skilllearnbench.js","sourceRoot":"","sources":["../../../src/eval/live/skilllearnbench.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAC1C,OAAO,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AACzC,OAAO,EAAE,UAAU,EAAa,MAAM,sBAAsB,CAAC;AAI7D,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAC5C,OAAO,EAAE,uBAAuB,EAAE,MAAM,sBAAsB,CAAC;AAC/D,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AAkCtC,MAAM,OAAO,wBAAwB;IAC1B,IAAI,GAAG,iBAAiB,CAAC;IACzB,WAAW,GAAG,KAAK,CAAC;IACpB,aAAa,GAAG,IAAI,CAAC;IACrB,uBAAuB,GAAG,CAAC,CAAC,CAAC,yDAAyD;IAEvF,MAAM,GAAqB,IAAI,CAAC;IACvB,UAAU,CAAS;IAEpC,YAAY,OAAgC,EAAE;QAC5C,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU,IAAI,CAAC,CAAC;IACzC,CAAC;IAED,KAAK,CAAC,KAAK;QACT,MAAM,IAAI,CAAC,SAAS,EAAE,CAAC;IACzB,CAAC;IAED,KAAK,CAAC,QAAQ;QACZ,MAAM,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE,CAAC;QAC3B,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC;IACrB,CAAC;IAEO,KAAK,CAAC,SAAS;QACrB,IAAI,IAAI,CAAC,MAAM;YAAE,OAAO,IAAI,CAAC,MAAM,CAAC;QACpC,MAAM,EAAE,OAAO,EAAE,IAAI,EAAE,GAAG,oBAAoB,EAAE,CAAC;QACjD,MAAM,MAAM,GAAG,IAAI,SAAS,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;QAC5C,IAAI,CAAC;YACH,MAAM,MAAM,CAAC,KAAK,EAAE,CAAC;QACvB,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,MAAM,IAAI,KAAK,CACb,2CAA4C,GAAa,CAAC,OAAO,MAAM;gBACrE,6EAA6E;gBAC7E,oEAAoE,CACvE,CAAC;QACJ,CAAC;QACD,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,OAAO,MAAM,CAAC;IAChB,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,SAAS,CAAC,IAAqC;QACnD,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,SAAS,EAAE,CAAC;QACtC,MAAM,QAAQ,GAAG,MAAM,MAAM,CAAC,IAAI,CAAkC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC;QAErF,+EAA+E;QAC/E,MAAM,SAAS,GAAG,CAAC,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC;QACtC,MAAM,GAAG,GAAG,SAAS,CAAC,2BAA2B,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC;QAC9D,KAAK,IAAI,CAAC,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;YAC9C,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACtC,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;QAC9D,CAAC;QAED,MAAM,SAAS,GAAe,EAAE,CAAC;QACjC,IAAI,KAAK,GAAG,CAAC,CAAC;QAEd,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YACjC,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,IAAI,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,CAAC;gBAClD,SAAS,CAAC,IAAI,CAAC;oBACb,IAAI,EAAE,UAAU,CAAC;wBACf,EAAE,EAAE,cAAc,QAAQ,CAAC,EAAE,QAAQ,IAAI,EAAE;wBAC3C,MAAM,EAAE,QAAQ,CAAC,MAAM;wBACvB,WAAW,EAAE,QAAQ,CAAC,WAAW;wBACjC,QAAQ,EAAE;4BACR,QAAQ,EAAE,QAAQ,CAAC,MAAM;4BACzB,MAAM,EAAE,QAAQ,CAAC,EAAE;4BACnB,SAAS,EAAE,IAAI;4BACf,eAAe,EAAE,IAAI,CAAC,UAAU;4BAChC,SAAS,EAAE,KAAK;yBACjB;qBACF,CAAC;oBACF,QAAQ,EAAE,QAAQ,CAAC,MAAM;oBACzB,KAAK;oBACL,QAAQ,EAAE,CAAC;oBACX,QAAQ,EAAE,CAAC;oBACX,MAAM,EAAE,QAAQ,CAAC,EAAE;oBACnB,SAAS,EAAE,IAAI;iBAChB,CAAC,CAAC;gBACH,KAAK,EAAE,CAAC;gBACR,IAAI,KAAK,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAM;YACjC,CAAC;YACD,IAAI,KAAK,IAAI,IAAI,CAAC,KAAK;gBAAE,MAAM;QACjC,CAAC;QAED,OAAO,SAAS,CAAC;IACnB,CAAC;IAED;;;OAGG;IACH,MAAM,CAAC,KAAW,EAAE,OAAe;QACjC,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC;IAC5B,CAAC;IAED,iBAAiB;QACf,OAAO,IAAI,CAAC,MAAM,CAAC;IACrB,CAAC;IAED,eAAe,GAAG,CAChB,UAAiC,EACjC,IAA6C,EAC1B,EAAE;QACrB,OAAO,IAAI,kBAAkB,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC5C,CAAC,CAAC;CACH;AAED,MAAM,UAAU,8BAA8B,CAAC,IAE9C;IACC,OAAO,IAAI,wBAAwB,CAAC,IAAI,CAAC,CAAC;AAC5C,CAAC;AAED;;;GAGG;AACH,MAAM,kBAAmB,SAAQ,uBAAuB;IAEnC;IADnB,YACmB,KAA+B,EAChD,IAA6C;QAE7C,KAAK,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC;QAHF,UAAK,GAAL,KAAK,CAA0B;IAIlD,CAAC;IAEQ,KAAK,CAAC,OAAO,CACpB,MAAc,EACd,OAA8D;QAE9D,MAAM,MAAM,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;QAEpD,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,iBAAiB,EAAE,CAAC;QAC9C,IAAI,MAAM,EAAE,CAAC;YACX,IAAI,CAAC;gBACH,MAAM,YAAY,GAAG,MAAM,MAAM,CAAC,IAAI,CAA2B;oBAC/D,GAAG,EAAE,QAAQ;oBACb,MAAM,EAAE,MAAM,CAAC,MAAM;iBACtB,CAAC,CAAC;gBACH,OAAO;oBACL,GAAG,MAAM;oBACT,OAAO,EAAE,YAAY,CAAC,YAAY;oBAClC,UAAU,EAAE;wBACV,GAAG,CAAE,MAAM,CAAC,UAAsC,IAAI,EAAE,CAAC;wBACzD,YAAY,EAAE,YAAY,CAAC,aAAa;wBACxC,iBAAiB,EAAE,YAAY,CAAC,kBAAkB;wBAClD,YAAY,EAAE,YAAY,CAAC,aAAa;qBACzC;iBACF,CAAC;YACJ,CAAC;YAAC,MAAM,CAAC;gBACP,uDAAuD;YACzD,CAAC;QACH,CAAC;QAED,OAAO,MAAM,CAAC;IAChB,CAAC;CACF;AAED,SAAS,oBAAoB;IAC3B,MAAM,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,sBAAsB,CAAC;IAC/C,IAAI,GAAG,IAAI,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC;QACtB,MAAM,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;QACtC,OAAO,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;IACrD,CAAC;IACD,MAAM,IAAI,GAAG,OAAO,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;IACrD,MAAM,MAAM,GAAG,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAChF,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,EAAE,SAAS,EAAE,2BAA2B,CAAC,CAAC;IACpE,OAAO,EAAE,OAAO,EAAE,OAAO,CAAC,GAAG,CAAC,MAAM,IAAI,SAAS,EAAE,IAAI,EAAE,CAAC,MAAM,CAAC,EAAE,CAAC;AACtE,CAAC"}
|
|
@@ -0,0 +1,96 @@
|
|
|
1
|
+
import { type GepaOperatorConfig } from '../../learning/gepa-proposer.js';
|
|
2
|
+
import type { SkillOptV1Memory } from './skillopt_v1_memory.js';
|
|
3
|
+
import type { SkillOptV1Causal } from './skillopt_v1_causal.js';
|
|
4
|
+
export { evaluateChecks, scoreRuleJudge, describeCheck, type RuleCheck, type RuleJudge, type CheckResult, type SkillOptV1Task, } from './skillopt_v1_rules.js';
|
|
5
|
+
import { type SkillOptV1Task } from './skillopt_v1_rules.js';
|
|
6
|
+
export interface SkillOptV1Seed {
|
|
7
|
+
name: string;
|
|
8
|
+
skill: string;
|
|
9
|
+
tasks: SkillOptV1Task[];
|
|
10
|
+
}
|
|
11
|
+
/**
|
|
12
|
+
* Load one seed. gbrain's held-out.jsonl → `test` (the final metric);
|
|
13
|
+
* benchmark.jsonl → split into `train` + `val` (every 3rd item → val). The
|
|
14
|
+
* train/val split only steers optimization; `test` is identical to the baselines,
|
|
15
|
+
* so the reported metric is comparable regardless of the split rule.
|
|
16
|
+
*/
|
|
17
|
+
export declare function loadSkillOptV1Seed(dataRoot: string, seed: string): SkillOptV1Seed;
|
|
18
|
+
export declare function availableSkillOptV1Seeds(): string[];
|
|
19
|
+
/** char/4 token accounting + prompt-hash cache, matching the baselines' CliBackend. */
|
|
20
|
+
export declare class TokenMeter {
|
|
21
|
+
private readonly bedrock;
|
|
22
|
+
private readonly awsRegion?;
|
|
23
|
+
tokens: number;
|
|
24
|
+
private cache;
|
|
25
|
+
constructor(bedrock?: boolean, awsRegion?: string | undefined);
|
|
26
|
+
private static key;
|
|
27
|
+
call(prompt: string, model: string): Promise<string>;
|
|
28
|
+
}
|
|
29
|
+
export interface SkillOptV1RunConfig {
|
|
30
|
+
dataRoot: string;
|
|
31
|
+
seed: string;
|
|
32
|
+
targetModel: string;
|
|
33
|
+
proposerModel: string;
|
|
34
|
+
/** GEPA metric-call budget. */
|
|
35
|
+
budget?: number;
|
|
36
|
+
/** cap the TRAIN reflection pool (cost control). */
|
|
37
|
+
limit?: number;
|
|
38
|
+
/**
|
|
39
|
+
* Fix 1 (val/test gap): cap val/test independently of `limit`. Default
|
|
40
|
+
* undefined = use ALL available (val 5, test 6 on skillopt-v1) so val-perfect is
|
|
41
|
+
* a reliable stop signal and the held-out metric isn't crippled to 3 tasks.
|
|
42
|
+
*/
|
|
43
|
+
valLimit?: number;
|
|
44
|
+
testLimit?: number;
|
|
45
|
+
/** seeded RNG for reproducibility. */
|
|
46
|
+
rng?: () => number;
|
|
47
|
+
/**
|
|
48
|
+
* Phase-2 lever: ccore memory. When supplied, optimization rollouts are
|
|
49
|
+
* recorded and the proposer is conditioned on retrieved successful exemplars
|
|
50
|
+
* (held-out rollouts are never recorded). Omit for the vanilla-GEPA parity
|
|
51
|
+
* baseline. A shared instance across seeds gives cross-seed accumulation.
|
|
52
|
+
*/
|
|
53
|
+
memory?: SkillOptV1Memory;
|
|
54
|
+
/** Operator-design toggles (bounded edits / gate / single-candidate) for the parity work. */
|
|
55
|
+
operator?: GepaOperatorConfig;
|
|
56
|
+
/** Use the stall-triggered escalation ladder instead of a single fixed operator. */
|
|
57
|
+
escalate?: boolean;
|
|
58
|
+
/** Fix 2: slow-update/meta-skill reflection at escalation tier boundaries. */
|
|
59
|
+
slowUpdate?: boolean;
|
|
60
|
+
/** Route LM calls through Bedrock (env-var path; needs ~/.aws creds). */
|
|
61
|
+
bedrock?: boolean;
|
|
62
|
+
awsRegion?: string;
|
|
63
|
+
/**
|
|
64
|
+
* Phase-2 lever 2: ccore causal store. When supplied, rollouts are recorded as
|
|
65
|
+
* contrastive cause→effect edges and the proposer is given the strongest
|
|
66
|
+
* discriminators. Composes with (and is independent of) `memory`.
|
|
67
|
+
*/
|
|
68
|
+
causal?: SkillOptV1Causal;
|
|
69
|
+
/** Raw memory-conditioning hook (advanced/testing). Ignored if `memory`/`causal` set. */
|
|
70
|
+
memoryContext?: (tasks: SkillOptV1Task[]) => Promise<string> | string;
|
|
71
|
+
}
|
|
72
|
+
export interface SkillOptV1Record {
|
|
73
|
+
arm: 'ccore';
|
|
74
|
+
config: string;
|
|
75
|
+
optimizer: string;
|
|
76
|
+
target: string;
|
|
77
|
+
seed: string;
|
|
78
|
+
held_out_before: number;
|
|
79
|
+
held_out_after: number;
|
|
80
|
+
'reached_1.0': boolean;
|
|
81
|
+
tokens: number;
|
|
82
|
+
metric_calls: number;
|
|
83
|
+
iterations: number;
|
|
84
|
+
val_best: number;
|
|
85
|
+
/** whether ccore memory-conditioning (lever 1) was active. */
|
|
86
|
+
memory: boolean;
|
|
87
|
+
/** whether ccore causal-contrast (lever 2) was active. */
|
|
88
|
+
causal: boolean;
|
|
89
|
+
/** operator mode: 'rewrite-none-frontier' (GEPA) … 'bounded-val-single' (parity); 'escalation'. */
|
|
90
|
+
operatorMode: string;
|
|
91
|
+
/** escalation only: the tier that produced the best (routing label). */
|
|
92
|
+
tierReached?: string;
|
|
93
|
+
}
|
|
94
|
+
/** Run ccore's proposer over one seed and score held-out. Mirrors run_gepa.py. */
|
|
95
|
+
export declare function runSkillOptV1(cfg: SkillOptV1RunConfig): Promise<SkillOptV1Record>;
|
|
96
|
+
//# sourceMappingURL=skillopt_v1.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skillopt_v1.d.ts","sourceRoot":"","sources":["../../../src/eval/live/skillopt_v1.ts"],"names":[],"mappings":"AAoBA,OAAO,EAGL,KAAK,kBAAkB,EACxB,MAAM,iCAAiC,CAAC;AAEzC,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAC;AAChE,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAC;AAKhE,OAAO,EACL,cAAc,EACd,cAAc,EACd,aAAa,EACb,KAAK,SAAS,EACd,KAAK,SAAS,EACd,KAAK,WAAW,EAChB,KAAK,cAAc,GACpB,MAAM,wBAAwB,CAAC;AAChC,OAAO,EAAkC,KAAK,cAAc,EAAE,MAAM,wBAAwB,CAAC;AAW7F,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,cAAc,EAAE,CAAC;CACzB;AA8BD;;;;;GAKG;AACH,wBAAgB,kBAAkB,CAAC,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,cAAc,CAWjF;AAED,wBAAgB,wBAAwB,IAAI,MAAM,EAAE,CAEnD;AAID,uFAAuF;AACvF,qBAAa,UAAU;IAKnB,OAAO,CAAC,QAAQ,CAAC,OAAO;IACxB,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC;IAL7B,MAAM,SAAK;IACX,OAAO,CAAC,KAAK,CAA6B;gBAGvB,OAAO,UAAQ,EACf,SAAS,CAAC,EAAE,MAAM,YAAA;IAGrC,OAAO,CAAC,MAAM,CAAC,GAAG;IAWZ,IAAI,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CAe3D;AAsBD,MAAM,WAAW,mBAAmB;IAClC,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,EAAE,MAAM,CAAC;IACpB,aAAa,EAAE,MAAM,CAAC;IACtB,+BAA+B;IAC/B,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,oDAAoD;IACpD,KAAK,CAAC,EAAE,MAAM,CAAC;IACf;;;;OAIG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,sCAAsC;IACtC,GAAG,CAAC,EAAE,MAAM,MAAM,CAAC;IACnB;;;;;OAKG;IACH,MAAM,CAAC,EAAE,gBAAgB,CAAC;IAC1B,6FAA6F;IAC7F,QAAQ,CAAC,EAAE,kBAAkB,CAAC;IAC9B,oFAAoF;IACpF,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,8EAA8E;IAC9E,UAAU,CAAC,EAAE,OAAO,CAAC;IACrB,yEAAyE;IACzE,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB;;;;OAIG;IACH,MAAM,CAAC,EAAE,gBAAgB,CAAC;IAC1B,yFAAyF;IACzF,aAAa,CAAC,EAAE,CAAC,KAAK,EAAE,cAAc,EAAE,KAAK,OAAO,CAAC,MAAM,CAAC,GAAG,MAAM,CAAC;CACvE;AAED,MAAM,WAAW,gBAAgB;IAC/B,GAAG,EAAE,OAAO,CAAC;IACb,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,MAAM,CAAC;IAClB,MAAM,EAAE,MAAM,CAAC;IACf,IAAI,EAAE,MAAM,CAAC;IACb,eAAe,EAAE,MAAM,CAAC;IACxB,cAAc,EAAE,MAAM,CAAC;IACvB,aAAa,EAAE,OAAO,CAAC;IACvB,MAAM,EAAE,MAAM,CAAC;IACf,YAAY,EAAE,MAAM,CAAC;IACrB,UAAU,EAAE,MAAM,CAAC;IACnB,QAAQ,EAAE,MAAM,CAAC;IACjB,8DAA8D;IAC9D,MAAM,EAAE,OAAO,CAAC;IAChB,0DAA0D;IAC1D,MAAM,EAAE,OAAO,CAAC;IAChB,mGAAmG;IACnG,YAAY,EAAE,MAAM,CAAC;IACrB,wEAAwE;IACxE,WAAW,CAAC,EAAE,MAAM,CAAC;CACtB;AAMD,kFAAkF;AAClF,wBAAsB,aAAa,CAAC,GAAG,EAAE,mBAAmB,GAAG,OAAO,CAAC,gBAAgB,CAAC,CA+IvF"}
|
|
Binary file
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skillopt_v1.js","sourceRoot":"","sources":["../../../src/eval/live/skillopt_v1.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AACH,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,SAAS,CAAC;AACnD,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAC5C,OAAO,EACL,eAAe,GAGhB,MAAM,iCAAiC,CAAC;AACzC,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAC;AAI7D,6EAA6E;AAC7E,6EAA6E;AAC7E,2CAA2C;AAC3C,OAAO,EACL,cAAc,EACd,cAAc,EACd,aAAa,GAKd,MAAM,wBAAwB,CAAC;AAChC,OAAO,EAAE,cAAc,EAAuC,MAAM,wBAAwB,CAAC;AAE7F,8EAA8E;AAE9E,MAAM,SAAS,GAA2B;IACxC,cAAc,EAAE,wBAAwB;IACxC,kBAAkB,EAAE,cAAc;IAClC,OAAO,EAAE,iBAAiB;IAC1B,gBAAgB,EAAE,qBAAqB;CACxC,CAAC;AAQF,SAAS,SAAS,CAAC,IAAY;IAC7B,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;QAAE,OAAO,EAAE,CAAC;IACjC,MAAM,GAAG,GAAmC,EAAE,CAAC;IAC/C,KAAK,MAAM,IAAI,IAAI,YAAY,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QAC1D,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QACtB,IAAI,CAAC,CAAC;YAAE,SAAS;QACjB,IAAI,CAAC;YACH,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;QAC1B,CAAC;QAAC,MAAM,CAAC;YACP,oBAAoB;QACtB,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,MAAM,CACb,GAA4B,EAC5B,IAAY,EACZ,KAA8B;IAE9B,OAAO;QACL,EAAE,EAAE,GAAG,IAAI,IAAI,MAAM,CAAC,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC,EAAE;QAC1C,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC;QAC9B,KAAK,EAAG,GAAG,CAAC,KAAmB,IAAI,EAAE;QACrC,KAAK;KACN,CAAC;AACJ,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,kBAAkB,CAAC,QAAgB,EAAE,IAAY;IAC/D,MAAM,GAAG,GAAG,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC;IACpD,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,CAAC,CAAC;QAC7C,CAAC,CAAC,YAAY,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,CAAC,EAAE,MAAM,CAAC;QAC7C,CAAC,CAAC,EAAE,CAAC;IACP,MAAM,KAAK,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,iBAAiB,CAAC,CAAC,CAAC;IACtD,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,gBAAgB,CAAC,CAAC,CAAC;IACpD,MAAM,KAAK,GAAqB,EAAE,CAAC;IACnC,KAAK,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;IACxF,IAAI,CAAC,OAAO,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC;IAC7D,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC;AACtC,CAAC;AAED,MAAM,UAAU,wBAAwB;IACtC,OAAO,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;AAChC,CAAC;AAED,8EAA8E;AAE9E,uFAAuF;AACvF,MAAM,OAAO,UAAU;IAKF;IACA;IALnB,MAAM,GAAG,CAAC,CAAC;IACH,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAC;IAE1C,YACmB,UAAU,KAAK,EACf,SAAkB;QADlB,YAAO,GAAP,OAAO,CAAQ;QACf,cAAS,GAAT,SAAS,CAAS;IAClC,CAAC;IAEI,MAAM,CAAC,GAAG,CAAC,KAAa,EAAE,MAAc;QAC9C,+CAA+C;QAC/C,IAAI,CAAC,GAAG,UAAU,CAAC;QACnB,MAAM,CAAC,GAAG,GAAG,KAAK,IAAI,MAAM,EAAE,CAAC;QAC/B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAClC,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;YACrB,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE,UAAU,CAAC,CAAC;QAC/B,CAAC;QACD,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC;IAChC,CAAC;IAED,KAAK,CAAC,IAAI,CAAC,MAAc,EAAE,KAAa;QACtC,MAAM,GAAG,GAAG,UAAU,CAAC,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;QAC1C,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,MAAM,KAAK,SAAS;YAAE,OAAO,MAAM,CAAC,CAAC,mCAAmC;QAC5E,MAAM,GAAG,GAAG,MAAM,SAAS,CAAC,MAAM,EAAE;YAClC,KAAK;YACL,aAAa,EAAE,IAAI;YACnB,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,SAAS,EAAE,IAAI,CAAC,SAAS;SAC1B,CAAC,CAAC;QACH,MAAM,GAAG,GAAG,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QACpC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAC1E,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;QACzB,OAAO,GAAG,CAAC;IACb,CAAC;CACF;AAED,8EAA8E;AAE9E,mEAAmE;AACnE,SAAS,aAAa,CAAC,KAAa,EAAE,MAAc;IAClD,OAAO;QACL,gFAAgF;QAChF,6EAA6E;QAC7E,gFAAgF;QAChF,sCAAsC;QACtC,EAAE;QACF,YAAY,KAAK,IAAI,QAAQ,EAAE;QAC/B,EAAE;QACF,WAAW,MAAM,EAAE;QACnB,EAAE;QACF,kDAAkD;KACnD,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACf,CAAC;AAuED,SAAS,GAAG,CAAI,EAAO,EAAE,CAAU;IACjC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;AAC1C,CAAC;AAED,kFAAkF;AAClF,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,GAAwB;IAC1D,MAAM,IAAI,GAAG,kBAAkB,CAAC,GAAG,CAAC,QAAQ,EAAE,GAAG,CAAC,IAAI,CAAC,CAAC;IACxD,MAAM,KAAK,GAAG,GAAG,CACf,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,KAAK,OAAO,CAAC,EAC7C,GAAG,CAAC,KAAK,CACV,CAAC;IACF,2EAA2E;IAC3E,2EAA2E;IAC3E,0EAA0E;IAC1E,IAAI,GAAG,GAAG,GAAG,CACX,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,KAAK,KAAK,CAAC,EAC3C,GAAG,CAAC,QAAQ,CACb,CAAC;IACF,IAAI,GAAG,CAAC,MAAM,KAAK,CAAC;QAAE,GAAG,GAAG,KAAK,CAAC;IAClC,MAAM,IAAI,GAAG,GAAG,CACd,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,KAAK,MAAM,CAAC,EAC5C,GAAG,CAAC,SAAS,CACd,CAAC;IAEF,MAAM,KAAK,GAAG,IAAI,UAAU,CAAC,GAAG,CAAC,OAAO,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC;IAEzD,gEAAgE;IAChE,MAAM,OAAO,GAAG,KAAK,EAAE,KAAa,EAAE,KAAuB,EAA2B,EAAE;QACxF,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAa,EAAE,CAAC;QAC7B,MAAM,QAAQ,GAAa,EAAE,CAAC;QAC9B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;YACtB,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,IAAI,CAAC,aAAa,CAAC,KAAK,EAAE,CAAC,CAAC,MAAM,CAAC,EAAE,GAAG,CAAC,WAAW,CAAC,CAAC;YAC9E,MAAM,CAAC,GAAG,cAAc,CAAC,CAAC,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;YACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;YACpB,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YAClB,QAAQ,CAAC,IAAI,CACX,CAAC,CAAC,IAAI,IAAI,CAAC;gBACT,CAAC,CAAC,IAAI;gBACN,CAAC,CAAC,GAAG,CAAC,CAAC,SAAS,6BAA6B,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE,CAAC,EAAE,CACtF,CAAC;QACJ,CAAC;QACD,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAC;IACvC,CAAC,CAAC;IAEF,gFAAgF;IAChF,MAAM,QAAQ,GAAG,KAAK,EAAE,KAAa,EAAE,KAAuB,EAA2B,EAAE;QACzF,MAAM,EAAE,GAAG,MAAM,OAAO,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;QACvC,IAAI,GAAG,CAAC,MAAM;YAAE,MAAM,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QACxD,IAAI,GAAG,CAAC,MAAM;YAAE,MAAM,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QACxD,OAAO,EAAE,CAAC;IACZ,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,CAAC,MAAc,EAAmB,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,CAAC,aAAa,CAAC,CAAC;IAE3F,uEAAuE;IACvE,MAAM,SAAS,GAAG,KAAK,EAAE,KAAa,EAAmB,EAAE;QACzD,MAAM,EAAE,GAAG,MAAM,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC;QACtC,OAAO,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;IACxF,CAAC,CAAC;IAEF,mFAAmF;IACnF,MAAM,SAAS,GAA6D,EAAE,CAAC;IAC/E,IAAI,GAAG,CAAC,MAAM;QAAE,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,MAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9D,IAAI,GAAG,CAAC,MAAM;QAAE,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,MAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9D,MAAM,aAAa,GAAG,SAAS,CAAC,MAAM;QACpC,CAAC,CAAC,KAAK,EAAE,KAAuB,EAAE,EAAE,CAChC,CAAC,MAAM,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC;QACpF,CAAC,CAAC,GAAG,CAAC,aAAa,CAAC;IAEtB,+EAA+E;IAC/E,MAAM,GAAG,GAAG,IAAI,GAAG,EAAU,CAAC;IAC9B,KAAK,MAAM,CAAC,IAAI,CAAC,GAAG,KAAK,EAAE,GAAG,GAAG,CAAC;QAAE,KAAK,MAAM,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE;YAAE,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;IACxF,MAAM,QAAQ,GAAG;QACf,IAAI,EAAE,GAAG,CAAC,IAAI;QACd,QAAQ,EAAE,CAAC,GAAG,GAAG,CAAC;QAClB,WAAW,EAAE,GAAG,CAAC,GAAG,CAAC,WAAW,CAAC;QACjC,iBAAiB,EAAE,GAAG,CAAC,GAAG,CAAC,iBAAiB,CAAC;KAC9C,CAAC;IAEF,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAC3C,MAAM,MAAM,GAAG;QACb,SAAS,EAAE,IAAI,CAAC,KAAK;QACrB,QAAQ,EAAE,KAAK;QACf,MAAM,EAAE,GAAG;QACX,QAAQ;QACR,OAAO;QACP,YAAY,EAAE,CAAC,CAAiB,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM;QAC7C,YAAY,EAAE,CAAC;QACf,uBAAuB,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC;QACvD,GAAG,EAAE,GAAG,CAAC,GAAG;QACZ,aAAa;KACd,CAAC;IAEF,IAAI,SAAiB,CAAC;IACtB,IAAI,YAAoB,CAAC;IACzB,IAAI,WAAmB,CAAC;IACxB,IAAI,UAAkB,CAAC;IACvB,IAAI,WAA+B,CAAC;IACpC,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC;QACjB,MAAM,CAAC,GAAG,MAAM,aAAa,CAAiB;YAC5C,GAAG,MAAM;YACT,cAAc,EAAE,GAAG,CAAC,MAAM,IAAI,EAAE;YAChC,QAAQ;YACR,UAAU,EAAE,GAAG,CAAC,UAAU;SAC3B,CAAC,CAAC;QACH,SAAS,GAAG,CAAC,CAAC,SAAS,CAAC;QACxB,YAAY,GAAG,CAAC,CAAC,YAAY,CAAC;QAC9B,WAAW,GAAG,CAAC,CAAC,WAAW,CAAC;QAC5B,UAAU,GAAG,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC;QAC9B,WAAW,GAAG,CAAC,CAAC,WAAW,CAAC;IAC9B,CAAC;SAAM,CAAC;QACN,MAAM,CAAC,GAAG,MAAM,eAAe,CAAiB;YAC9C,GAAG,MAAM;YACT,cAAc,EAAE,GAAG,CAAC,MAAM,IAAI,EAAE;YAChC,QAAQ,EAAE,GAAG,CAAC,QAAQ;SACvB,CAAC,CAAC;QACH,SAAS,GAAG,CAAC,CAAC,SAAS,CAAC;QACxB,YAAY,GAAG,CAAC,CAAC,YAAY,CAAC;QAC9B,WAAW,GAAG,CAAC,CAAC,WAAW,CAAC;QAC5B,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC;IAC5B,CAAC;IACD,MAAM,KAAK,GAAG,MAAM,SAAS,CAAC,SAAS,CAAC,CAAC;IAEzC,OAAO;QACL,GAAG,EAAE,OAAO;QACZ,MAAM,EAAE,GAAG,CAAC,aAAa,KAAK,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ;QACjE,SAAS,EAAE,GAAG,CAAC,aAAa;QAC5B,MAAM,EAAE,GAAG,CAAC,WAAW;QACvB,IAAI,EAAE,GAAG,CAAC,IAAI;QACd,eAAe,EAAE,MAAM,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;QAC1C,cAAc,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;QACxC,aAAa,EAAE,KAAK,IAAI,KAAK;QAC7B,MAAM,EAAE,KAAK,CAAC,MAAM;QACpB,YAAY,EAAE,WAAW;QACzB,UAAU;QACV,QAAQ,EAAE,MAAM,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;QACzC,MAAM,EAAE,CAAC,CAAC,GAAG,CAAC,MAAM;QACpB,MAAM,EAAE,CAAC,CAAC,GAAG,CAAC,MAAM;QACpB,YAAY,EAAE,GAAG,CAAC,QAAQ;YACxB,CAAC,CAAC,YAAY;YACd,CAAC,CAAC;gBACE,GAAG,CAAC,QAAQ,EAAE,QAAQ,IAAI,SAAS;gBACnC,GAAG,CAAC,QAAQ,EAAE,IAAI,IAAI,MAAM;gBAC5B,GAAG,CAAC,QAAQ,EAAE,SAAS,IAAI,UAAU;aACtC,CAAC,IAAI,CAAC,GAAG,CAAC;QACf,WAAW;KACZ,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,26 @@
|
|
|
1
|
+
import type { GepaEvalResult } from '../../learning/gepa-proposer.js';
|
|
2
|
+
import { type SkillOptV1Task } from './skillopt_v1_rules.js';
|
|
3
|
+
export interface SkillOptV1CausalConfig {
|
|
4
|
+
domain?: string;
|
|
5
|
+
/** how many causal edges to surface in the reflection prompt (default 5). */
|
|
6
|
+
k?: number;
|
|
7
|
+
/** min observedCount before an edge is surfaced (noise floor, default 1). */
|
|
8
|
+
minObserved?: number;
|
|
9
|
+
}
|
|
10
|
+
export declare class SkillOptV1Causal {
|
|
11
|
+
private readonly store;
|
|
12
|
+
private readonly domain;
|
|
13
|
+
private readonly k;
|
|
14
|
+
private readonly minObserved;
|
|
15
|
+
private counter;
|
|
16
|
+
private constructor();
|
|
17
|
+
static create(cfg?: SkillOptV1CausalConfig): Promise<SkillOptV1Causal>;
|
|
18
|
+
private edge;
|
|
19
|
+
/** Record contrastive edges from one rollout. */
|
|
20
|
+
record(task: SkillOptV1Task, output: string, score: number): Promise<void>;
|
|
21
|
+
recordBatch(tasks: SkillOptV1Task[], ev: GepaEvalResult): Promise<void>;
|
|
22
|
+
/** Render the strongest discriminating causal edges for the minibatch. */
|
|
23
|
+
context(tasks: SkillOptV1Task[]): Promise<string>;
|
|
24
|
+
close(): Promise<void>;
|
|
25
|
+
}
|
|
26
|
+
//# sourceMappingURL=skillopt_v1_causal.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skillopt_v1_causal.d.ts","sourceRoot":"","sources":["../../../src/eval/live/skillopt_v1_causal.ts"],"names":[],"mappings":"AAmBA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACtE,OAAO,EAAkB,KAAK,cAAc,EAAE,MAAM,wBAAwB,CAAC;AAK7E,MAAM,WAAW,sBAAsB;IACrC,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,6EAA6E;IAC7E,CAAC,CAAC,EAAE,MAAM,CAAC;IACX,6EAA6E;IAC7E,WAAW,CAAC,EAAE,MAAM,CAAC;CACtB;AAED,qBAAa,gBAAgB;IAC3B,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAc;IACpC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAS;IAC3B,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAS;IACrC,OAAO,CAAC,OAAO,CAAK;IAEpB,OAAO;WAOM,MAAM,CAAC,GAAG,GAAE,sBAA2B,GAAG,OAAO,CAAC,gBAAgB,CAAC;YASlE,IAAI;IAgBlB,iDAAiD;IAC3C,MAAM,CAAC,IAAI,EAAE,cAAc,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAgB1E,WAAW,CAAC,KAAK,EAAE,cAAc,EAAE,EAAE,EAAE,EAAE,cAAc,GAAG,OAAO,CAAC,IAAI,CAAC;IAM7E,0EAA0E;IACpE,OAAO,CAAC,KAAK,EAAE,cAAc,EAAE,GAAG,OAAO,CAAC,MAAM,CAAC;IAoBjD,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;CAG7B"}
|
|
@@ -0,0 +1,106 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Phase-2 lever 2 (reflexion + causal reflection inputs) for the skillopt-v1
|
|
3
|
+
* head-to-head — backed by ccore's real `CausalStore`.
|
|
4
|
+
*
|
|
5
|
+
* Distinct from lever 1 (retrieval of *successful* exemplars → "winners had X"):
|
|
6
|
+
* this records CONTRASTIVE cause→effect edges from BOTH passes and failures —
|
|
7
|
+
* include "<structural feature>" → passes all checks
|
|
8
|
+
* omit "<required element>" → fails the <op> check
|
|
9
|
+
* — and surfaces the strongest discriminators, naming exactly what the current
|
|
10
|
+
* failing answers lack. Whether this adds value over lever 1 on a structural
|
|
11
|
+
* benchmark is an empirical (ablation) question; it is toggled independently.
|
|
12
|
+
*
|
|
13
|
+
* See docs/DESIGN-skillopt-headtohead.md §4 / §9c.
|
|
14
|
+
*/
|
|
15
|
+
import { tmpdir } from 'node:os';
|
|
16
|
+
import { join } from 'node:path';
|
|
17
|
+
import { createSqlitePersistence } from '../../persistence/index.js';
|
|
18
|
+
import { CausalStore } from '../../experimental/causal-store.js';
|
|
19
|
+
import { evaluateChecks } from './skillopt_v1_rules.js';
|
|
20
|
+
import { structuralSignature } from './skillopt_v1_memory.js';
|
|
21
|
+
const PASS_EFFECT = 'passes all checks';
|
|
22
|
+
export class SkillOptV1Causal {
|
|
23
|
+
store;
|
|
24
|
+
domain;
|
|
25
|
+
k;
|
|
26
|
+
minObserved;
|
|
27
|
+
counter = 0;
|
|
28
|
+
constructor(store, cfg) {
|
|
29
|
+
this.store = store;
|
|
30
|
+
this.domain = cfg.domain ?? 'skillopt-v1';
|
|
31
|
+
this.k = cfg.k ?? 5;
|
|
32
|
+
this.minObserved = cfg.minObserved ?? 1;
|
|
33
|
+
}
|
|
34
|
+
static async create(cfg = {}) {
|
|
35
|
+
const baseDir = join(tmpdir(), `skillopt_v1_causal_${process.pid}_${Date.now()}`);
|
|
36
|
+
const persistence = createSqlitePersistence({ baseDir });
|
|
37
|
+
await persistence.init();
|
|
38
|
+
const c = new SkillOptV1Causal(new CausalStore(persistence), cfg);
|
|
39
|
+
await c.store.init();
|
|
40
|
+
return c;
|
|
41
|
+
}
|
|
42
|
+
async edge(cause, effect, tid) {
|
|
43
|
+
const now = new Date();
|
|
44
|
+
const e = {
|
|
45
|
+
id: `ce-${this.domain}-${this.counter++}`,
|
|
46
|
+
cause,
|
|
47
|
+
effect,
|
|
48
|
+
confidence: 0.5,
|
|
49
|
+
domain: this.domain,
|
|
50
|
+
sourceTrajectoryIds: [tid],
|
|
51
|
+
observedCount: 1,
|
|
52
|
+
createdAt: now,
|
|
53
|
+
updatedAt: now,
|
|
54
|
+
};
|
|
55
|
+
await this.store.add(e); // merges with an existing match (bumps observedCount)
|
|
56
|
+
}
|
|
57
|
+
/** Record contrastive edges from one rollout. */
|
|
58
|
+
async record(task, output, score) {
|
|
59
|
+
const tid = `${task.id}#${this.counter}`;
|
|
60
|
+
const checks = evaluateChecks(task.judge, output);
|
|
61
|
+
if (score >= 1) {
|
|
62
|
+
// a full pass — the structural features present plausibly caused it
|
|
63
|
+
for (const f of structuralSignature(output)) {
|
|
64
|
+
await this.edge(`include ${f}`, PASS_EFFECT, tid);
|
|
65
|
+
}
|
|
66
|
+
}
|
|
67
|
+
else {
|
|
68
|
+
// attribute each failed check to the missing requirement
|
|
69
|
+
for (const c of checks) {
|
|
70
|
+
if (!c.pass)
|
|
71
|
+
await this.edge(`omit ${c.desc}`, `fails the ${c.op} check`, tid);
|
|
72
|
+
}
|
|
73
|
+
}
|
|
74
|
+
}
|
|
75
|
+
async recordBatch(tasks, ev) {
|
|
76
|
+
for (let i = 0; i < tasks.length; i++) {
|
|
77
|
+
await this.record(tasks[i], ev.outputs[i] ?? '', ev.scores[i] ?? 0);
|
|
78
|
+
}
|
|
79
|
+
}
|
|
80
|
+
/** Render the strongest discriminating causal edges for the minibatch. */
|
|
81
|
+
async context(tasks) {
|
|
82
|
+
const seen = new Set();
|
|
83
|
+
const edges = [];
|
|
84
|
+
for (const t of tasks) {
|
|
85
|
+
for (const e of await this.store.findSimilar(t.intent, { k: this.k })) {
|
|
86
|
+
if (seen.has(e.id) || e.observedCount < this.minObserved)
|
|
87
|
+
continue;
|
|
88
|
+
seen.add(e.id);
|
|
89
|
+
edges.push(e);
|
|
90
|
+
}
|
|
91
|
+
}
|
|
92
|
+
if (edges.length === 0)
|
|
93
|
+
return '';
|
|
94
|
+
// strongest first: most-observed, then most-confident
|
|
95
|
+
edges.sort((a, b) => b.observedCount - a.observedCount || b.confidence - a.confidence);
|
|
96
|
+
const top = edges.slice(0, this.k);
|
|
97
|
+
return [
|
|
98
|
+
'Causal signals from your own rollouts (fix the failures, keep the passes):',
|
|
99
|
+
...top.map((e) => `- ${e.cause} → ${e.effect} (seen ${e.observedCount}x)`),
|
|
100
|
+
].join('\n');
|
|
101
|
+
}
|
|
102
|
+
async close() {
|
|
103
|
+
// CausalStore has no close(); the ephemeral SQLite dir is GC'd with the process.
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
//# sourceMappingURL=skillopt_v1_causal.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skillopt_v1_causal.js","sourceRoot":"","sources":["../../../src/eval/live/skillopt_v1_causal.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AACH,OAAO,EAAE,MAAM,EAAE,MAAM,SAAS,CAAC;AACjC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,uBAAuB,EAAE,MAAM,4BAA4B,CAAC;AACrE,OAAO,EAAE,WAAW,EAAE,MAAM,oCAAoC,CAAC;AAGjE,OAAO,EAAE,cAAc,EAAuB,MAAM,wBAAwB,CAAC;AAC7E,OAAO,EAAE,mBAAmB,EAAE,MAAM,yBAAyB,CAAC;AAE9D,MAAM,WAAW,GAAG,mBAAmB,CAAC;AAUxC,MAAM,OAAO,gBAAgB;IACV,KAAK,CAAc;IACnB,MAAM,CAAS;IACf,CAAC,CAAS;IACV,WAAW,CAAS;IAC7B,OAAO,GAAG,CAAC,CAAC;IAEpB,YAAoB,KAAkB,EAAE,GAA2B;QACjE,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,MAAM,GAAG,GAAG,CAAC,MAAM,IAAI,aAAa,CAAC;QAC1C,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;QACpB,IAAI,CAAC,WAAW,GAAG,GAAG,CAAC,WAAW,IAAI,CAAC,CAAC;IAC1C,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,MAA8B,EAAE;QAClD,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,EAAE,EAAE,sBAAsB,OAAO,CAAC,GAAG,IAAI,IAAI,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QAClF,MAAM,WAAW,GAAG,uBAAuB,CAAC,EAAE,OAAO,EAAE,CAAC,CAAC;QACzD,MAAM,WAAW,CAAC,IAAI,EAAE,CAAC;QACzB,MAAM,CAAC,GAAG,IAAI,gBAAgB,CAAC,IAAI,WAAW,CAAC,WAAW,CAAC,EAAE,GAAG,CAAC,CAAC;QAClE,MAAM,CAAC,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QACrB,OAAO,CAAC,CAAC;IACX,CAAC;IAEO,KAAK,CAAC,IAAI,CAAC,KAAa,EAAE,MAAc,EAAE,GAAW;QAC3D,MAAM,GAAG,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,MAAM,CAAC,GAAe;YACpB,EAAE,EAAE,MAAM,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,OAAO,EAAE,EAAE;YACzC,KAAK;YACL,MAAM;YACN,UAAU,EAAE,GAAG;YACf,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,mBAAmB,EAAE,CAAC,GAAG,CAAC;YAC1B,aAAa,EAAE,CAAC;YAChB,SAAS,EAAE,GAAG;YACd,SAAS,EAAE,GAAG;SACf,CAAC;QACF,MAAM,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,sDAAsD;IACjF,CAAC;IAED,iDAAiD;IACjD,KAAK,CAAC,MAAM,CAAC,IAAoB,EAAE,MAAc,EAAE,KAAa;QAC9D,MAAM,GAAG,GAAG,GAAG,IAAI,CAAC,EAAE,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;QACzC,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;QAClD,IAAI,KAAK,IAAI,CAAC,EAAE,CAAC;YACf,oEAAoE;YACpE,KAAK,MAAM,CAAC,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC5C,MAAM,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,EAAE,EAAE,WAAW,EAAE,GAAG,CAAC,CAAC;YACpD,CAAC;QACH,CAAC;aAAM,CAAC;YACN,yDAAyD;YACzD,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;gBACvB,IAAI,CAAC,CAAC,CAAC,IAAI;oBAAE,MAAM,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,EAAE,aAAa,CAAC,CAAC,EAAE,QAAQ,EAAE,GAAG,CAAC,CAAC;YACjF,CAAC;QACH,CAAC;IACH,CAAC;IAED,KAAK,CAAC,WAAW,CAAC,KAAuB,EAAE,EAAkB;QAC3D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACtC,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACtE,CAAC;IACH,CAAC;IAED,0EAA0E;IAC1E,KAAK,CAAC,OAAO,CAAC,KAAuB;QACnC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;QAC/B,MAAM,KAAK,GAAiB,EAAE,CAAC;QAC/B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;YACtB,KAAK,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,EAAE,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC;gBACtE,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,aAAa,GAAG,IAAI,CAAC,WAAW;oBAAE,SAAS;gBACnE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;gBACf,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YAChB,CAAC;QACH,CAAC;QACD,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QAClC,sDAAsD;QACtD,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,aAAa,GAAG,CAAC,CAAC,aAAa,IAAI,CAAC,CAAC,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC;QACvF,MAAM,GAAG,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC;QACnC,OAAO;YACL,4EAA4E;YAC5E,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC,KAAK,MAAM,CAAC,CAAC,MAAM,UAAU,CAAC,CAAC,aAAa,IAAI,CAAC;SAC3E,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACf,CAAC;IAED,KAAK,CAAC,KAAK;QACT,iFAAiF;IACnF,CAAC;CACF"}
|
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
import type { GepaEvalResult } from '../../learning/gepa-proposer.js';
|
|
2
|
+
import type { SkillOptV1Task } from './skillopt_v1_rules.js';
|
|
3
|
+
export interface SkillOptV1MemoryConfig {
|
|
4
|
+
/** retrieval depth (ReMem default 4). */
|
|
5
|
+
k?: number;
|
|
6
|
+
/** domain partition — keep one so all seeds share a pool (cross-seed transfer). */
|
|
7
|
+
domain?: string;
|
|
8
|
+
/** only surface successful exemplars (default true). */
|
|
9
|
+
successOnly?: boolean;
|
|
10
|
+
/**
|
|
11
|
+
* How retrieved exemplars are rendered into the reflection prompt:
|
|
12
|
+
* - 'recipe' (default): a compact union of the STRUCTURAL signature (headings,
|
|
13
|
+
* bold labels, `Label:` lines) the winning answers shared — ~30-60 tokens,
|
|
14
|
+
* exactly what the structural judge checks. ~15-25x cheaper than 'excerpt'.
|
|
15
|
+
* - 'excerpt': the raw winning-answer text (verbose; the original Phase-2 build).
|
|
16
|
+
* Falls back to a short excerpt when an exemplar has no extractable structure
|
|
17
|
+
* (e.g. winners that only satisfy a max_chars/contains check).
|
|
18
|
+
*/
|
|
19
|
+
mode?: 'recipe' | 'excerpt';
|
|
20
|
+
/** max distinct exemplars folded in (dedup cap, default 6). */
|
|
21
|
+
maxExemplars?: number;
|
|
22
|
+
/** max chars per exemplar in 'excerpt' mode / recipe fallback (default 240). */
|
|
23
|
+
excerptChars?: number;
|
|
24
|
+
}
|
|
25
|
+
/**
|
|
26
|
+
* Extract the structural skeleton of an answer: markdown headings, bold labels,
|
|
27
|
+
* and `Label:` lines — the elements section_present / regex checks look for.
|
|
28
|
+
*/
|
|
29
|
+
export declare function structuralSignature(answer: string): string[];
|
|
30
|
+
export declare class SkillOptV1Memory {
|
|
31
|
+
private readonly memory;
|
|
32
|
+
private readonly k;
|
|
33
|
+
private readonly domain;
|
|
34
|
+
private readonly successOnly;
|
|
35
|
+
private readonly mode;
|
|
36
|
+
private readonly maxExemplars;
|
|
37
|
+
private readonly excerptChars;
|
|
38
|
+
private counter;
|
|
39
|
+
private constructor();
|
|
40
|
+
/** Build an isolated, ephemeral memory (own temp SQLite dir). */
|
|
41
|
+
static create(cfg?: SkillOptV1MemoryConfig): Promise<SkillOptV1Memory>;
|
|
42
|
+
/** Record one optimization rollout as an experience (success keyed on hard score). */
|
|
43
|
+
record(task: SkillOptV1Task, output: string, score: number, feedback: string): Promise<void>;
|
|
44
|
+
/** Record a whole evaluated batch (parallel arrays). */
|
|
45
|
+
recordBatch(tasks: SkillOptV1Task[], ev: GepaEvalResult): Promise<void>;
|
|
46
|
+
/**
|
|
47
|
+
* Retrieve top-k similar successful exemplars for a minibatch and render them
|
|
48
|
+
* as reflection context. Deduplicates by experience id across the batch.
|
|
49
|
+
*/
|
|
50
|
+
context(tasks: SkillOptV1Task[]): Promise<string>;
|
|
51
|
+
close(): Promise<void>;
|
|
52
|
+
}
|
|
53
|
+
//# sourceMappingURL=skillopt_v1_memory.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"skillopt_v1_memory.d.ts","sourceRoot":"","sources":["../../../src/eval/live/skillopt_v1_memory.ts"],"names":[],"mappings":"AAsBA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACtE,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,wBAAwB,CAAC;AAE7D,MAAM,WAAW,sBAAsB;IACrC,yCAAyC;IACzC,CAAC,CAAC,EAAE,MAAM,CAAC;IACX,mFAAmF;IACnF,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,wDAAwD;IACxD,WAAW,CAAC,EAAE,OAAO,CAAC;IACtB;;;;;;;;OAQG;IACH,IAAI,CAAC,EAAE,QAAQ,GAAG,SAAS,CAAC;IAC5B,+DAA+D;IAC/D,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,gFAAgF;IAChF,YAAY,CAAC,EAAE,MAAM,CAAC;CACvB;AAED;;;GAGG;AACH,wBAAgB,mBAAmB,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,EAAE,CAW5D;AAED,qBAAa,gBAAgB;IAC3B,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAmB;IAC1C,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAS;IAC3B,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAU;IACtC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAuB;IAC5C,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAS;IACtC,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAS;IACtC,OAAO,CAAC,OAAO,CAAK;IAEpB,OAAO;IAUP,iEAAiE;WACpD,MAAM,CAAC,GAAG,GAAE,sBAA2B,GAAG,OAAO,CAAC,gBAAgB,CAAC;IAShF,sFAAsF;IAChF,MAAM,CACV,IAAI,EAAE,cAAc,EACpB,MAAM,EAAE,MAAM,EACd,KAAK,EAAE,MAAM,EACb,QAAQ,EAAE,MAAM,GACf,OAAO,CAAC,IAAI,CAAC;IAoBhB,wDAAwD;IAClD,WAAW,CAAC,KAAK,EAAE,cAAc,EAAE,EAAE,EAAE,EAAE,cAAc,GAAG,OAAO,CAAC,IAAI,CAAC;IAM7E;;;OAGG;IACG,OAAO,CAAC,KAAK,EAAE,cAAc,EAAE,GAAG,OAAO,CAAC,MAAM,CAAC;IA6CjD,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;CAG7B"}
|