@namzu/sdk 30.2.0 → 31.1.0
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/CHANGELOG.md +562 -0
- package/README.md +43 -2
- package/dist/advisory/executor.d.ts +2 -1
- package/dist/advisory/executor.d.ts.map +1 -1
- package/dist/advisory/executor.js +4 -1
- package/dist/advisory/executor.js.map +1 -1
- package/dist/agents/ReactiveAgent.d.ts.map +1 -1
- package/dist/agents/ReactiveAgent.js +12 -0
- package/dist/agents/ReactiveAgent.js.map +1 -1
- package/dist/agents/RouterAgent.d.ts.map +1 -1
- package/dist/agents/RouterAgent.js +21 -3
- package/dist/agents/RouterAgent.js.map +1 -1
- package/dist/agents/SupervisorAgent.d.ts.map +1 -1
- package/dist/agents/SupervisorAgent.js +34 -11
- package/dist/agents/SupervisorAgent.js.map +1 -1
- package/dist/agents/runAgent.d.ts +12 -0
- package/dist/agents/runAgent.d.ts.map +1 -1
- package/dist/agents/runAgent.js +26 -2
- package/dist/agents/runAgent.js.map +1 -1
- package/dist/authorization/command-line.d.ts +81 -0
- package/dist/authorization/command-line.d.ts.map +1 -0
- package/dist/authorization/command-line.js +347 -0
- package/dist/authorization/command-line.js.map +1 -0
- package/dist/authorization/rules.d.ts.map +1 -1
- package/dist/authorization/rules.js +22 -1
- package/dist/authorization/rules.js.map +1 -1
- package/dist/bridge/a2a/client.d.ts +23 -5
- package/dist/bridge/a2a/client.d.ts.map +1 -1
- package/dist/bridge/a2a/client.js +320 -73
- package/dist/bridge/a2a/client.js.map +1 -1
- package/dist/bridge/a2a/mapper.d.ts.map +1 -1
- package/dist/bridge/a2a/mapper.js +4 -0
- package/dist/bridge/a2a/mapper.js.map +1 -1
- package/dist/bridge/acp/server.d.ts +13 -1
- package/dist/bridge/acp/server.d.ts.map +1 -1
- package/dist/bridge/acp/server.js +150 -44
- package/dist/bridge/acp/server.js.map +1 -1
- package/dist/bridge/mcp/connector/adapter.d.ts +3 -2
- package/dist/bridge/mcp/connector/adapter.d.ts.map +1 -1
- package/dist/bridge/mcp/connector/adapter.js +17 -21
- package/dist/bridge/mcp/connector/adapter.js.map +1 -1
- package/dist/bridge/sse/mapper.d.ts.map +1 -1
- package/dist/bridge/sse/mapper.js +3 -0
- package/dist/bridge/sse/mapper.js.map +1 -1
- package/dist/compaction/dangling.d.ts +29 -2
- package/dist/compaction/dangling.d.ts.map +1 -1
- package/dist/compaction/dangling.js +172 -35
- package/dist/compaction/dangling.js.map +1 -1
- package/dist/compaction/index.d.ts +3 -2
- package/dist/compaction/index.d.ts.map +1 -1
- package/dist/compaction/index.js +1 -1
- package/dist/compaction/index.js.map +1 -1
- package/dist/compaction/manual.d.ts +11 -1
- package/dist/compaction/manual.d.ts.map +1 -1
- package/dist/compaction/manual.js +89 -14
- package/dist/compaction/manual.js.map +1 -1
- package/dist/compaction/plan.d.ts +19 -6
- package/dist/compaction/plan.d.ts.map +1 -1
- package/dist/compaction/plan.js +18 -4
- package/dist/compaction/plan.js.map +1 -1
- package/dist/compaction/retention.d.ts +10 -7
- package/dist/compaction/retention.d.ts.map +1 -1
- package/dist/compaction/retention.js +26 -8
- package/dist/compaction/retention.js.map +1 -1
- package/dist/compaction/summary.d.ts +4 -4
- package/dist/compaction/summary.js +4 -4
- package/dist/compaction/verifier.d.ts +19 -1
- package/dist/compaction/verifier.d.ts.map +1 -1
- package/dist/compaction/verifier.js +20 -1
- package/dist/compaction/verifier.js.map +1 -1
- package/dist/config/runtime.d.ts +6 -6
- package/dist/connector/BaseConnector.d.ts +6 -4
- package/dist/connector/BaseConnector.d.ts.map +1 -1
- package/dist/connector/BaseConnector.js +8 -2
- package/dist/connector/BaseConnector.js.map +1 -1
- package/dist/connector/builtins/http.d.ts +16 -4
- package/dist/connector/builtins/http.d.ts.map +1 -1
- package/dist/connector/builtins/http.js +177 -44
- package/dist/connector/builtins/http.js.map +1 -1
- package/dist/connector/builtins/webhook.d.ts +15 -4
- package/dist/connector/builtins/webhook.d.ts.map +1 -1
- package/dist/connector/builtins/webhook.js +165 -45
- package/dist/connector/builtins/webhook.js.map +1 -1
- package/dist/connector/execution-contract.d.ts +11 -0
- package/dist/connector/execution-contract.d.ts.map +1 -0
- package/dist/connector/execution-contract.js +14 -0
- package/dist/connector/execution-contract.js.map +1 -0
- package/dist/connector/http-operation.d.ts +48 -0
- package/dist/connector/http-operation.d.ts.map +1 -0
- package/dist/connector/http-operation.js +205 -0
- package/dist/connector/http-operation.js.map +1 -0
- package/dist/connector/mcp/adapter.d.ts.map +1 -1
- package/dist/connector/mcp/adapter.js +8 -3
- package/dist/connector/mcp/adapter.js.map +1 -1
- package/dist/connector/mcp/client.d.ts +14 -8
- package/dist/connector/mcp/client.d.ts.map +1 -1
- package/dist/connector/mcp/client.js +160 -44
- package/dist/connector/mcp/client.js.map +1 -1
- package/dist/connector/mcp/http-sse.d.ts +8 -2
- package/dist/connector/mcp/http-sse.d.ts.map +1 -1
- package/dist/connector/mcp/http-sse.js +70 -17
- package/dist/connector/mcp/http-sse.js.map +1 -1
- package/dist/connector/mcp/prompt-adapter.d.ts.map +1 -1
- package/dist/connector/mcp/prompt-adapter.js +9 -2
- package/dist/connector/mcp/prompt-adapter.js.map +1 -1
- package/dist/connector/mcp/stdio.d.ts +2 -2
- package/dist/connector/mcp/stdio.d.ts.map +1 -1
- package/dist/connector/mcp/stdio.js +14 -50
- package/dist/connector/mcp/stdio.js.map +1 -1
- package/dist/connector/mcp/streamable-http.d.ts +7 -2
- package/dist/connector/mcp/streamable-http.d.ts.map +1 -1
- package/dist/connector/mcp/streamable-http.js +66 -18
- package/dist/connector/mcp/streamable-http.js.map +1 -1
- package/dist/connector/tools/adapter.d.ts +2 -2
- package/dist/connector/tools/adapter.d.ts.map +1 -1
- package/dist/connector/tools/adapter.js +23 -14
- package/dist/connector/tools/adapter.js.map +1 -1
- package/dist/connector/tools/definitions.d.ts.map +1 -1
- package/dist/connector/tools/definitions.js +5 -2
- package/dist/connector/tools/definitions.js.map +1 -1
- package/dist/connector/tools/result.d.ts +4 -0
- package/dist/connector/tools/result.d.ts.map +1 -0
- package/dist/connector/tools/result.js +20 -0
- package/dist/connector/tools/result.js.map +1 -0
- package/dist/connector/web/guarded-fetch.d.ts +4 -2
- package/dist/connector/web/guarded-fetch.d.ts.map +1 -1
- package/dist/connector/web/guarded-fetch.js +157 -43
- package/dist/connector/web/guarded-fetch.js.map +1 -1
- package/dist/constants/sandbox/index.d.ts +3 -2
- package/dist/constants/sandbox/index.d.ts.map +1 -1
- package/dist/constants/sandbox/index.js +3 -2
- package/dist/constants/sandbox/index.js.map +1 -1
- package/dist/contracts/api.d.ts +2 -0
- package/dist/contracts/api.d.ts.map +1 -1
- package/dist/contracts/schemas.d.ts +38 -12
- package/dist/contracts/schemas.d.ts.map +1 -1
- package/dist/contracts/schemas.js +7 -0
- package/dist/contracts/schemas.js.map +1 -1
- package/dist/directory/derive-supervisor.d.ts.map +1 -1
- package/dist/directory/derive-supervisor.js +8 -0
- package/dist/directory/derive-supervisor.js.map +1 -1
- package/dist/directory/derive.d.ts.map +1 -1
- package/dist/directory/derive.js +8 -0
- package/dist/directory/derive.js.map +1 -1
- package/dist/directory/load.d.ts.map +1 -1
- package/dist/directory/load.js +2 -0
- package/dist/directory/load.js.map +1 -1
- package/dist/directory/types.d.ts +4 -0
- package/dist/directory/types.d.ts.map +1 -1
- package/dist/eval/experiment.d.ts +10 -10
- package/dist/eval/experiment.d.ts.map +1 -1
- package/dist/eval/experiment.js +121 -91
- package/dist/eval/experiment.js.map +1 -1
- package/dist/eval/judge.d.ts +6 -0
- package/dist/eval/judge.d.ts.map +1 -1
- package/dist/eval/judge.js +7 -2
- package/dist/eval/judge.js.map +1 -1
- package/dist/eval/types.d.ts +7 -1
- package/dist/eval/types.d.ts.map +1 -1
- package/dist/manager/agent/lifecycle.d.ts +12 -10
- package/dist/manager/agent/lifecycle.d.ts.map +1 -1
- package/dist/manager/agent/lifecycle.js +16 -3
- package/dist/manager/agent/lifecycle.js.map +1 -1
- package/dist/manager/connector/environment.d.ts +2 -2
- package/dist/manager/connector/environment.d.ts.map +1 -1
- package/dist/manager/connector/environment.js +2 -2
- package/dist/manager/connector/environment.js.map +1 -1
- package/dist/manager/connector/lifecycle.d.ts +10 -2
- package/dist/manager/connector/lifecycle.d.ts.map +1 -1
- package/dist/manager/connector/lifecycle.js +273 -24
- package/dist/manager/connector/lifecycle.js.map +1 -1
- package/dist/manager/connector/tenant.d.ts +5 -4
- package/dist/manager/connector/tenant.d.ts.map +1 -1
- package/dist/manager/connector/tenant.js +47 -22
- package/dist/manager/connector/tenant.js.map +1 -1
- package/dist/manager/goal/activation.d.ts +23 -0
- package/dist/manager/goal/activation.d.ts.map +1 -0
- package/dist/manager/goal/activation.js +41 -0
- package/dist/manager/goal/activation.js.map +1 -0
- package/dist/manager/index.d.ts +3 -2
- package/dist/manager/index.d.ts.map +1 -1
- package/dist/manager/index.js.map +1 -1
- package/dist/manager/run/persistence.d.ts +2 -0
- package/dist/manager/run/persistence.d.ts.map +1 -1
- package/dist/manager/run/persistence.js +5 -1
- package/dist/manager/run/persistence.js.map +1 -1
- package/dist/manager/topic/dependency.d.ts +24 -0
- package/dist/manager/topic/dependency.d.ts.map +1 -0
- package/dist/manager/topic/dependency.js +13 -0
- package/dist/manager/topic/dependency.js.map +1 -0
- package/dist/manager/topic/lifecycle.d.ts +9 -21
- package/dist/manager/topic/lifecycle.d.ts.map +1 -1
- package/dist/manager/topic/lifecycle.js +17 -29
- package/dist/manager/topic/lifecycle.js.map +1 -1
- package/dist/plugin/index.d.ts +1 -0
- package/dist/plugin/index.d.ts.map +1 -1
- package/dist/plugin/index.js.map +1 -1
- package/dist/plugin/lifecycle.d.ts +15 -0
- package/dist/plugin/lifecycle.d.ts.map +1 -1
- package/dist/plugin/lifecycle.js +144 -38
- package/dist/plugin/lifecycle.js.map +1 -1
- package/dist/plugin/loader.d.ts +3 -3
- package/dist/plugin/loader.d.ts.map +1 -1
- package/dist/plugin/loader.js +58 -12
- package/dist/plugin/loader.js.map +1 -1
- package/dist/provider/collect-chat-completion.d.ts.map +1 -1
- package/dist/provider/collect-chat-completion.js +4 -0
- package/dist/provider/collect-chat-completion.js.map +1 -1
- package/dist/provider/fallback.d.ts +3 -1
- package/dist/provider/fallback.d.ts.map +1 -1
- package/dist/provider/fallback.js +62 -10
- package/dist/provider/fallback.js.map +1 -1
- package/dist/provider/idle-timeout.d.ts +4 -0
- package/dist/provider/idle-timeout.d.ts.map +1 -1
- package/dist/provider/idle-timeout.js +131 -15
- package/dist/provider/idle-timeout.js.map +1 -1
- package/dist/provider/index.d.ts +2 -0
- package/dist/provider/index.d.ts.map +1 -1
- package/dist/provider/index.js +1 -0
- package/dist/provider/index.js.map +1 -1
- package/dist/provider/instrumentation.d.ts.map +1 -1
- package/dist/provider/instrumentation.js +1 -0
- package/dist/provider/instrumentation.js.map +1 -1
- package/dist/provider/retry.d.ts.map +1 -1
- package/dist/provider/retry.js +13 -1
- package/dist/provider/retry.js.map +1 -1
- package/dist/public-runtime.d.ts +57 -24
- package/dist/public-runtime.d.ts.map +1 -1
- package/dist/public-runtime.js +42 -35
- package/dist/public-runtime.js.map +1 -1
- package/dist/public-tools.d.ts +1 -0
- package/dist/public-tools.d.ts.map +1 -1
- package/dist/public-tools.js +5 -0
- package/dist/public-tools.js.map +1 -1
- package/dist/public-types.d.ts +33 -12
- package/dist/public-types.d.ts.map +1 -1
- package/dist/rag/embedding.d.ts +8 -3
- package/dist/rag/embedding.d.ts.map +1 -1
- package/dist/rag/embedding.js +120 -9
- package/dist/rag/embedding.js.map +1 -1
- package/dist/rag/index.d.ts +1 -1
- package/dist/rag/index.d.ts.map +1 -1
- package/dist/rag/index.js +1 -1
- package/dist/rag/index.js.map +1 -1
- package/dist/rag/ingestion.d.ts +2 -2
- package/dist/rag/ingestion.d.ts.map +1 -1
- package/dist/rag/ingestion.js +15 -3
- package/dist/rag/ingestion.js.map +1 -1
- package/dist/rag/knowledge-base.d.ts +3 -3
- package/dist/rag/knowledge-base.d.ts.map +1 -1
- package/dist/rag/knowledge-base.js +8 -4
- package/dist/rag/knowledge-base.js.map +1 -1
- package/dist/rag/operation.d.ts +10 -0
- package/dist/rag/operation.d.ts.map +1 -0
- package/dist/rag/operation.js +26 -0
- package/dist/rag/operation.js.map +1 -0
- package/dist/rag/rag-tool.d.ts.map +1 -1
- package/dist/rag/rag-tool.js +6 -3
- package/dist/rag/rag-tool.js.map +1 -1
- package/dist/rag/retriever.d.ts +2 -2
- package/dist/rag/retriever.d.ts.map +1 -1
- package/dist/rag/retriever.js +31 -15
- package/dist/rag/retriever.js.map +1 -1
- package/dist/rag/vector-store.d.ts +3 -3
- package/dist/rag/vector-store.d.ts.map +1 -1
- package/dist/rag/vector-store.js +4 -2
- package/dist/rag/vector-store.js.map +1 -1
- package/dist/registry/command/kernel-commands.d.ts +16 -4
- package/dist/registry/command/kernel-commands.d.ts.map +1 -1
- package/dist/registry/command/kernel-commands.js +171 -0
- package/dist/registry/command/kernel-commands.js.map +1 -1
- package/dist/run/reporter.d.ts.map +1 -1
- package/dist/run/reporter.js +9 -0
- package/dist/run/reporter.js.map +1 -1
- package/dist/run-query/index.d.ts +15 -1
- package/dist/run-query/index.d.ts.map +1 -1
- package/dist/run-query/index.js +69 -15
- package/dist/run-query/index.js.map +1 -1
- package/dist/runtime/query/context.d.ts.map +1 -1
- package/dist/runtime/query/context.js +10 -3
- package/dist/runtime/query/context.js.map +1 -1
- package/dist/runtime/query/executor.d.ts +3 -0
- package/dist/runtime/query/executor.d.ts.map +1 -1
- package/dist/runtime/query/executor.js +80 -18
- package/dist/runtime/query/executor.js.map +1 -1
- package/dist/runtime/query/guard.d.ts +9 -0
- package/dist/runtime/query/guard.d.ts.map +1 -1
- package/dist/runtime/query/guard.js +11 -0
- package/dist/runtime/query/guard.js.map +1 -1
- package/dist/runtime/query/index.d.ts +30 -1
- package/dist/runtime/query/index.d.ts.map +1 -1
- package/dist/runtime/query/index.js +548 -95
- package/dist/runtime/query/index.js.map +1 -1
- package/dist/runtime/query/iteration/index.d.ts.map +1 -1
- package/dist/runtime/query/iteration/index.js +74 -14
- package/dist/runtime/query/iteration/index.js.map +1 -1
- package/dist/runtime/query/iteration/phases/compaction.d.ts.map +1 -1
- package/dist/runtime/query/iteration/phases/compaction.js +94 -53
- package/dist/runtime/query/iteration/phases/compaction.js.map +1 -1
- package/dist/runtime/query/iteration/phases/context.d.ts +3 -0
- package/dist/runtime/query/iteration/phases/context.d.ts.map +1 -1
- package/dist/runtime/query/iteration/phases/context.js.map +1 -1
- package/dist/runtime/query/iteration/phases/tool-review.d.ts.map +1 -1
- package/dist/runtime/query/iteration/phases/tool-review.js +21 -0
- package/dist/runtime/query/iteration/phases/tool-review.js.map +1 -1
- package/dist/runtime/query/iteration/stream-turn.d.ts.map +1 -1
- package/dist/runtime/query/iteration/stream-turn.js +4 -0
- package/dist/runtime/query/iteration/stream-turn.js.map +1 -1
- package/dist/runtime/query/project-instructions.d.ts +69 -0
- package/dist/runtime/query/project-instructions.d.ts.map +1 -0
- package/dist/runtime/query/project-instructions.js +137 -0
- package/dist/runtime/query/project-instructions.js.map +1 -0
- package/dist/runtime/query/request-rich-content.d.ts +22 -0
- package/dist/runtime/query/request-rich-content.d.ts.map +1 -0
- package/dist/runtime/query/request-rich-content.js +122 -0
- package/dist/runtime/query/request-rich-content.js.map +1 -0
- package/dist/runtime/query/result.d.ts.map +1 -1
- package/dist/runtime/query/result.js +12 -1
- package/dist/runtime/query/result.js.map +1 -1
- package/dist/runtime/query/resume-pending.d.ts.map +1 -1
- package/dist/runtime/query/resume-pending.js +19 -20
- package/dist/runtime/query/resume-pending.js.map +1 -1
- package/dist/runtime/query/resume-run.d.ts.map +1 -1
- package/dist/runtime/query/resume-run.js +54 -8
- package/dist/runtime/query/resume-run.js.map +1 -1
- package/dist/runtime/query/run-state.d.ts +5 -0
- package/dist/runtime/query/run-state.d.ts.map +1 -1
- package/dist/runtime/query/run-state.js +6 -1
- package/dist/runtime/query/run-state.js.map +1 -1
- package/dist/runtime/query/sandbox-lifecycle.d.ts +54 -0
- package/dist/runtime/query/sandbox-lifecycle.d.ts.map +1 -0
- package/dist/runtime/query/sandbox-lifecycle.js +167 -0
- package/dist/runtime/query/sandbox-lifecycle.js.map +1 -0
- package/dist/sandbox/isolation.d.ts.map +1 -1
- package/dist/sandbox/isolation.js +9 -0
- package/dist/sandbox/isolation.js.map +1 -1
- package/dist/sandbox/provider/local.d.ts +56 -6
- package/dist/sandbox/provider/local.d.ts.map +1 -1
- package/dist/sandbox/provider/local.js +181 -76
- package/dist/sandbox/provider/local.js.map +1 -1
- package/dist/sandbox/terminal.d.ts +13 -5
- package/dist/sandbox/terminal.d.ts.map +1 -1
- package/dist/sandbox/terminal.js +11 -3
- package/dist/sandbox/terminal.js.map +1 -1
- package/dist/session/errors.d.ts +28 -29
- package/dist/session/errors.d.ts.map +1 -1
- package/dist/session/errors.js +28 -29
- package/dist/session/errors.js.map +1 -1
- package/dist/session/handoff/broadcast.d.ts +5 -8
- package/dist/session/handoff/broadcast.d.ts.map +1 -1
- package/dist/session/handoff/broadcast.js +16 -6
- package/dist/session/handoff/broadcast.js.map +1 -1
- package/dist/session/handoff/single.d.ts +5 -8
- package/dist/session/handoff/single.d.ts.map +1 -1
- package/dist/session/handoff/single.js +7 -4
- package/dist/session/handoff/single.js.map +1 -1
- package/dist/store/attachment/index.d.ts +11 -3
- package/dist/store/attachment/index.d.ts.map +1 -1
- package/dist/store/attachment/index.js +49 -4
- package/dist/store/attachment/index.js.map +1 -1
- package/dist/store/feedback/conformance.d.ts.map +1 -1
- package/dist/store/feedback/conformance.js +32 -0
- package/dist/store/feedback/conformance.js.map +1 -1
- package/dist/store/feedback/disk.d.ts +8 -6
- package/dist/store/feedback/disk.d.ts.map +1 -1
- package/dist/store/feedback/disk.js +111 -76
- package/dist/store/feedback/disk.js.map +1 -1
- package/dist/store/goal/index.d.ts +111 -0
- package/dist/store/goal/index.d.ts.map +1 -0
- package/dist/store/goal/index.js +436 -0
- package/dist/store/goal/index.js.map +1 -0
- package/dist/store/index.d.ts +1 -1
- package/dist/store/index.d.ts.map +1 -1
- package/dist/store/index.js +1 -1
- package/dist/store/index.js.map +1 -1
- package/dist/store/kv/revision-record-store.d.ts +70 -0
- package/dist/store/kv/revision-record-store.d.ts.map +1 -0
- package/dist/store/kv/revision-record-store.js +290 -0
- package/dist/store/kv/revision-record-store.js.map +1 -0
- package/dist/store/memory/disk.d.ts +9 -7
- package/dist/store/memory/disk.d.ts.map +1 -1
- package/dist/store/memory/disk.js +256 -98
- package/dist/store/memory/disk.js.map +1 -1
- package/dist/store/run/disk.d.ts +10 -2
- package/dist/store/run/disk.d.ts.map +1 -1
- package/dist/store/run/disk.js +70 -4
- package/dist/store/run/disk.js.map +1 -1
- package/dist/store/run/memory.d.ts +4 -2
- package/dist/store/run/memory.d.ts.map +1 -1
- package/dist/store/run/memory.js +15 -1
- package/dist/store/run/memory.js.map +1 -1
- package/dist/store/session/disk.d.ts +9 -0
- package/dist/store/session/disk.d.ts.map +1 -1
- package/dist/store/session/disk.js +57 -5
- package/dist/store/session/disk.js.map +1 -1
- package/dist/store/session/memory.d.ts +1 -0
- package/dist/store/session/memory.d.ts.map +1 -1
- package/dist/store/session/memory.js +14 -0
- package/dist/store/session/memory.js.map +1 -1
- package/dist/store/topic/memory.d.ts +1 -2
- package/dist/store/topic/memory.d.ts.map +1 -1
- package/dist/store/topic/memory.js +4 -5
- package/dist/store/topic/memory.js.map +1 -1
- package/dist/store/topic/objective.d.ts +8 -4
- package/dist/store/topic/objective.d.ts.map +1 -1
- package/dist/store/topic/objective.js +185 -100
- package/dist/store/topic/objective.js.map +1 -1
- package/dist/store/topic/state.d.ts +1 -2
- package/dist/store/topic/state.d.ts.map +1 -1
- package/dist/store/topic/state.js +50 -62
- package/dist/store/topic/state.js.map +1 -1
- package/dist/tools/builtins/bash.d.ts.map +1 -1
- package/dist/tools/builtins/bash.js +20 -0
- package/dist/tools/builtins/bash.js.map +1 -1
- package/dist/tools/builtins/computer-use.d.ts +9 -1
- package/dist/tools/builtins/computer-use.d.ts.map +1 -1
- package/dist/tools/builtins/computer-use.js.map +1 -1
- package/dist/tools/builtins/content-fingerprint.d.ts +8 -7
- package/dist/tools/builtins/content-fingerprint.d.ts.map +1 -1
- package/dist/tools/builtins/content-fingerprint.js +9 -10
- package/dist/tools/builtins/content-fingerprint.js.map +1 -1
- package/dist/tools/builtins/edit.d.ts +6 -1
- package/dist/tools/builtins/edit.d.ts.map +1 -1
- package/dist/tools/builtins/edit.js +166 -27
- package/dist/tools/builtins/edit.js.map +1 -1
- package/dist/tools/builtins/write-file.d.ts.map +1 -1
- package/dist/tools/builtins/write-file.js +45 -19
- package/dist/tools/builtins/write-file.js.map +1 -1
- package/dist/tools/command-arguments.d.ts +43 -0
- package/dist/tools/command-arguments.d.ts.map +1 -0
- package/dist/tools/command-arguments.js +52 -0
- package/dist/tools/command-arguments.js.map +1 -0
- package/dist/tools/coordinator/index.d.ts +2 -0
- package/dist/tools/coordinator/index.d.ts.map +1 -1
- package/dist/tools/coordinator/index.js +3 -1
- package/dist/tools/coordinator/index.js.map +1 -1
- package/dist/tools/defineTool.d.ts +10 -0
- package/dist/tools/defineTool.d.ts.map +1 -1
- package/dist/tools/defineTool.js +1 -0
- package/dist/tools/defineTool.js.map +1 -1
- package/dist/tools/env-scrub.d.ts +7 -5
- package/dist/tools/env-scrub.d.ts.map +1 -1
- package/dist/tools/env-scrub.js +7 -5
- package/dist/tools/env-scrub.js.map +1 -1
- package/dist/tools/goal/index.d.ts +20 -0
- package/dist/tools/goal/index.d.ts.map +1 -0
- package/dist/tools/goal/index.js +104 -0
- package/dist/tools/goal/index.js.map +1 -0
- package/dist/tools/memory/index.d.ts +16 -0
- package/dist/tools/memory/index.d.ts.map +1 -1
- package/dist/tools/memory/index.js +6 -2
- package/dist/tools/memory/index.js.map +1 -1
- package/dist/tools/memory/search.d.ts +4 -1
- package/dist/tools/memory/search.d.ts.map +1 -1
- package/dist/tools/memory/search.js +10 -2
- package/dist/tools/memory/search.js.map +1 -1
- package/dist/types/agent/base.d.ts +11 -3
- package/dist/types/agent/base.d.ts.map +1 -1
- package/dist/types/agent/factory.d.ts +2 -0
- package/dist/types/agent/factory.d.ts.map +1 -1
- package/dist/types/agent/reactive.d.ts +2 -0
- package/dist/types/agent/reactive.d.ts.map +1 -1
- package/dist/types/agent/supervisor.d.ts +6 -0
- package/dist/types/agent/supervisor.d.ts.map +1 -1
- package/dist/types/agent/task.d.ts +6 -2
- package/dist/types/agent/task.d.ts.map +1 -1
- package/dist/types/connector/core.d.ts +13 -8
- package/dist/types/connector/core.d.ts.map +1 -1
- package/dist/types/connector/definition.d.ts +37 -18
- package/dist/types/connector/definition.d.ts.map +1 -1
- package/dist/types/connector/http.d.ts +6 -0
- package/dist/types/connector/http.d.ts.map +1 -1
- package/dist/types/connector/mcp.d.ts +24 -6
- package/dist/types/connector/mcp.d.ts.map +1 -1
- package/dist/types/connector/tenant.d.ts +17 -7
- package/dist/types/connector/tenant.d.ts.map +1 -1
- package/dist/types/connector/webhook.d.ts +6 -0
- package/dist/types/connector/webhook.d.ts.map +1 -1
- package/dist/types/goal/index.d.ts +46 -0
- package/dist/types/goal/index.d.ts.map +1 -0
- package/dist/types/goal/index.js +2 -0
- package/dist/types/goal/index.js.map +1 -0
- package/dist/types/ids/index.d.ts +2 -0
- package/dist/types/ids/index.d.ts.map +1 -1
- package/dist/types/ids/index.js.map +1 -1
- package/dist/types/message/index.d.ts +79 -13
- package/dist/types/message/index.d.ts.map +1 -1
- package/dist/types/message/index.js +49 -3
- package/dist/types/message/index.js.map +1 -1
- package/dist/types/plugin/index.d.ts +7 -5
- package/dist/types/plugin/index.d.ts.map +1 -1
- package/dist/types/plugin/index.js.map +1 -1
- package/dist/types/provider/chat.d.ts +27 -4
- package/dist/types/provider/chat.d.ts.map +1 -1
- package/dist/types/provider/index.d.ts +1 -1
- package/dist/types/provider/index.d.ts.map +1 -1
- package/dist/types/provider/interface.d.ts +18 -7
- package/dist/types/provider/interface.d.ts.map +1 -1
- package/dist/types/provider/model.d.ts +12 -0
- package/dist/types/provider/model.d.ts.map +1 -1
- package/dist/types/provider/stream.d.ts +6 -0
- package/dist/types/provider/stream.d.ts.map +1 -1
- package/dist/types/rag/embedding.d.ts +15 -2
- package/dist/types/rag/embedding.d.ts.map +1 -1
- package/dist/types/rag/ingestion.d.ts +2 -2
- package/dist/types/rag/ingestion.d.ts.map +1 -1
- package/dist/types/rag/knowledge-base.d.ts +3 -3
- package/dist/types/rag/knowledge-base.d.ts.map +1 -1
- package/dist/types/rag/retrieval.d.ts +2 -2
- package/dist/types/rag/retrieval.d.ts.map +1 -1
- package/dist/types/rag/scope.d.ts +4 -0
- package/dist/types/rag/scope.d.ts.map +1 -1
- package/dist/types/rag/vector.d.ts +5 -2
- package/dist/types/rag/vector.d.ts.map +1 -1
- package/dist/types/run/config.d.ts +20 -0
- package/dist/types/run/config.d.ts.map +1 -1
- package/dist/types/run/events.d.ts +25 -1
- package/dist/types/run/events.d.ts.map +1 -1
- package/dist/types/run/events.js.map +1 -1
- package/dist/types/run/metadata.d.ts +2 -0
- package/dist/types/run/metadata.d.ts.map +1 -1
- package/dist/types/run/step.d.ts +2 -6
- package/dist/types/run/step.d.ts.map +1 -1
- package/dist/types/run/step.js.map +1 -1
- package/dist/types/run/store.d.ts +55 -2
- package/dist/types/run/store.d.ts.map +1 -1
- package/dist/types/sandbox/index.d.ts +47 -6
- package/dist/types/sandbox/index.d.ts.map +1 -1
- package/dist/types/sandbox/index.js +23 -10
- package/dist/types/sandbox/index.js.map +1 -1
- package/dist/types/session/store.d.ts +13 -0
- package/dist/types/session/store.d.ts.map +1 -1
- package/dist/types/tool/index.d.ts +33 -11
- package/dist/types/tool/index.d.ts.map +1 -1
- package/dist/types/tool/index.js.map +1 -1
- package/dist/types/topic/entity.d.ts +3 -4
- package/dist/types/topic/entity.d.ts.map +1 -1
- package/dist/types/topic/store.d.ts +3 -3
- package/dist/utils/id.d.ts +3 -1
- package/dist/utils/id.d.ts.map +1 -1
- package/dist/utils/id.js +4 -0
- package/dist/utils/id.js.map +1 -1
- package/dist/utils/process-environment.d.ts +26 -0
- package/dist/utils/process-environment.d.ts.map +1 -0
- package/dist/utils/process-environment.js +78 -0
- package/dist/utils/process-environment.js.map +1 -0
- package/dist/vault/InMemoryCredentialVault.d.ts +2 -0
- package/dist/vault/InMemoryCredentialVault.d.ts.map +1 -1
- package/dist/vault/InMemoryCredentialVault.js +24 -4
- package/dist/vault/InMemoryCredentialVault.js.map +1 -1
- package/dist/vault/instrumentation.d.ts.map +1 -1
- package/dist/vault/instrumentation.js +15 -0
- package/dist/vault/instrumentation.js.map +1 -1
- package/package.json +1 -1
- package/src/advisory/executor.ts +4 -1
- package/src/agents/ReactiveAgent.ts +12 -0
- package/src/agents/RouterAgent.ts +25 -3
- package/src/agents/SupervisorAgent.ts +33 -11
- package/src/agents/runAgent.ts +38 -2
- package/src/authorization/command-line.ts +390 -0
- package/src/authorization/rules.ts +22 -1
- package/src/bridge/a2a/client.ts +384 -97
- package/src/bridge/a2a/mapper.ts +5 -0
- package/src/bridge/acp/server.ts +184 -42
- package/src/bridge/mcp/connector/adapter.ts +28 -20
- package/src/bridge/sse/mapper.ts +3 -0
- package/src/compaction/dangling.ts +203 -38
- package/src/compaction/index.ts +9 -2
- package/src/compaction/manual.ts +125 -18
- package/src/compaction/plan.ts +42 -11
- package/src/compaction/retention.ts +23 -8
- package/src/compaction/summary.ts +4 -4
- package/src/compaction/verifier.ts +66 -0
- package/src/connector/BaseConnector.ts +19 -4
- package/src/connector/builtins/http.ts +236 -46
- package/src/connector/builtins/webhook.ts +207 -47
- package/src/connector/execution-contract.ts +27 -0
- package/src/connector/http-operation.ts +238 -0
- package/src/connector/mcp/adapter.ts +8 -3
- package/src/connector/mcp/client.ts +207 -58
- package/src/connector/mcp/http-sse.ts +91 -23
- package/src/connector/mcp/prompt-adapter.ts +8 -2
- package/src/connector/mcp/stdio.ts +21 -46
- package/src/connector/mcp/streamable-http.ts +86 -22
- package/src/connector/tools/adapter.ts +23 -14
- package/src/connector/tools/definitions.ts +5 -2
- package/src/connector/tools/result.ts +21 -0
- package/src/connector/web/guarded-fetch.ts +188 -48
- package/src/constants/sandbox/index.ts +3 -2
- package/src/contracts/api.ts +2 -0
- package/src/contracts/schemas.ts +7 -0
- package/src/directory/derive-supervisor.ts +8 -0
- package/src/directory/derive.ts +8 -0
- package/src/directory/load.ts +2 -0
- package/src/directory/types.ts +4 -0
- package/src/eval/experiment.ts +160 -111
- package/src/eval/judge.ts +13 -2
- package/src/eval/types.ts +7 -1
- package/src/manager/agent/lifecycle.ts +28 -12
- package/src/manager/connector/environment.ts +3 -1
- package/src/manager/connector/lifecycle.ts +364 -24
- package/src/manager/connector/tenant.ts +56 -23
- package/src/manager/goal/activation.ts +52 -0
- package/src/manager/index.ts +10 -2
- package/src/manager/run/persistence.ts +6 -1
- package/src/manager/topic/dependency.ts +39 -0
- package/src/manager/topic/lifecycle.ts +19 -31
- package/src/plugin/index.ts +1 -0
- package/src/plugin/lifecycle.ts +180 -39
- package/src/plugin/loader.ts +70 -13
- package/src/provider/collect-chat-completion.ts +3 -0
- package/src/provider/fallback.ts +79 -11
- package/src/provider/idle-timeout.ts +143 -17
- package/src/provider/index.ts +2 -0
- package/src/provider/instrumentation.ts +1 -0
- package/src/provider/retry.ts +16 -1
- package/src/public-runtime.ts +190 -32
- package/src/public-tools.ts +5 -0
- package/src/public-types.ts +85 -11
- package/src/rag/embedding.ts +147 -19
- package/src/rag/index.ts +1 -1
- package/src/rag/ingestion.ts +16 -2
- package/src/rag/knowledge-base.ts +13 -4
- package/src/rag/operation.ts +28 -0
- package/src/rag/rag-tool.ts +6 -3
- package/src/rag/retriever.ts +62 -11
- package/src/rag/vector-store.ts +8 -2
- package/src/registry/command/kernel-commands.ts +223 -4
- package/src/run/reporter.ts +10 -0
- package/src/run-query/index.ts +84 -15
- package/src/runtime/query/context.ts +13 -3
- package/src/runtime/query/executor.ts +90 -17
- package/src/runtime/query/guard.ts +12 -0
- package/src/runtime/query/index.ts +652 -103
- package/src/runtime/query/iteration/index.ts +89 -13
- package/src/runtime/query/iteration/phases/compaction.ts +120 -55
- package/src/runtime/query/iteration/phases/context.ts +4 -0
- package/src/runtime/query/iteration/phases/tool-review.ts +28 -0
- package/src/runtime/query/iteration/stream-turn.ts +3 -0
- package/src/runtime/query/project-instructions.ts +212 -0
- package/src/runtime/query/request-rich-content.ts +153 -0
- package/src/runtime/query/result.ts +12 -1
- package/src/runtime/query/resume-pending.ts +20 -18
- package/src/runtime/query/resume-run.ts +52 -8
- package/src/runtime/query/run-state.ts +11 -1
- package/src/runtime/query/sandbox-lifecycle.ts +210 -0
- package/src/sandbox/isolation.ts +9 -0
- package/src/sandbox/provider/local.ts +255 -90
- package/src/sandbox/terminal.ts +13 -5
- package/src/session/errors.ts +68 -42
- package/src/session/handoff/broadcast.ts +20 -14
- package/src/session/handoff/single.ts +14 -13
- package/src/store/attachment/index.ts +65 -3
- package/src/store/feedback/conformance.ts +39 -0
- package/src/store/feedback/disk.ts +128 -82
- package/src/store/goal/index.ts +594 -0
- package/src/store/index.ts +1 -1
- package/src/store/kv/revision-record-store.ts +346 -0
- package/src/store/memory/disk.ts +288 -113
- package/src/store/run/disk.ts +96 -5
- package/src/store/run/memory.ts +22 -2
- package/src/store/session/disk.ts +93 -6
- package/src/store/session/memory.ts +22 -0
- package/src/store/topic/memory.ts +4 -5
- package/src/store/topic/objective.ts +215 -100
- package/src/store/topic/state.ts +100 -63
- package/src/tools/builtins/bash.ts +21 -0
- package/src/tools/builtins/computer-use.ts +9 -1
- package/src/tools/builtins/content-fingerprint.ts +9 -12
- package/src/tools/builtins/edit.ts +190 -27
- package/src/tools/builtins/write-file.ts +55 -19
- package/src/tools/command-arguments.ts +54 -0
- package/src/tools/coordinator/index.ts +4 -1
- package/src/tools/defineTool.ts +11 -0
- package/src/tools/env-scrub.ts +7 -5
- package/src/tools/goal/index.ts +129 -0
- package/src/tools/memory/index.ts +26 -3
- package/src/tools/memory/search.ts +20 -3
- package/src/types/agent/base.ts +12 -3
- package/src/types/agent/factory.ts +2 -0
- package/src/types/agent/reactive.ts +2 -0
- package/src/types/agent/supervisor.ts +6 -0
- package/src/types/agent/task.ts +6 -2
- package/src/types/connector/core.ts +13 -8
- package/src/types/connector/definition.ts +54 -20
- package/src/types/connector/http.ts +7 -0
- package/src/types/connector/mcp.ts +27 -6
- package/src/types/connector/tenant.ts +21 -7
- package/src/types/connector/webhook.ts +7 -0
- package/src/types/goal/index.ts +50 -0
- package/src/types/ids/index.ts +2 -0
- package/src/types/message/index.ts +147 -12
- package/src/types/plugin/index.ts +7 -5
- package/src/types/provider/chat.ts +35 -4
- package/src/types/provider/index.ts +1 -1
- package/src/types/provider/interface.ts +22 -7
- package/src/types/provider/model.ts +13 -0
- package/src/types/provider/stream.ts +6 -0
- package/src/types/rag/embedding.ts +16 -2
- package/src/types/rag/ingestion.ts +2 -1
- package/src/types/rag/knowledge-base.ts +7 -3
- package/src/types/rag/retrieval.ts +2 -1
- package/src/types/rag/scope.ts +5 -0
- package/src/types/rag/vector.ts +5 -2
- package/src/types/run/config.ts +20 -0
- package/src/types/run/events.ts +28 -3
- package/src/types/run/metadata.ts +2 -0
- package/src/types/run/step.ts +2 -7
- package/src/types/run/store.ts +60 -2
- package/src/types/sandbox/index.ts +57 -14
- package/src/types/session/store.ts +18 -0
- package/src/types/tool/index.ts +33 -11
- package/src/types/topic/entity.ts +3 -4
- package/src/types/topic/store.ts +3 -3
- package/src/utils/id.ts +6 -0
- package/src/utils/process-environment.ts +101 -0
- package/src/vault/InMemoryCredentialVault.ts +29 -4
- package/src/vault/instrumentation.ts +24 -0
package/dist/eval/experiment.js
CHANGED
|
@@ -1,4 +1,5 @@
|
|
|
1
1
|
import { describeUncertainty, uncertaintyOf } from './uncertainty.js';
|
|
2
|
+
const MAX_CASE_TIMEOUT_MS = 2_147_483_647;
|
|
2
3
|
/**
|
|
3
4
|
* Run a dataset and score it.
|
|
4
5
|
*
|
|
@@ -15,6 +16,7 @@ export async function runExperiment(config) {
|
|
|
15
16
|
const startedAt = Date.now();
|
|
16
17
|
const threshold = config.passThreshold ?? 1;
|
|
17
18
|
const concurrency = Math.max(1, config.concurrency ?? 1);
|
|
19
|
+
const timeoutMs = validateCaseTimeoutMs(config.timeoutMs);
|
|
18
20
|
const results = new Array(config.cases.length);
|
|
19
21
|
let cursor = 0;
|
|
20
22
|
const worker = async () => {
|
|
@@ -23,65 +25,7 @@ export async function runExperiment(config) {
|
|
|
23
25
|
const evalCase = config.cases[index];
|
|
24
26
|
if (!evalCase)
|
|
25
27
|
return;
|
|
26
|
-
const
|
|
27
|
-
const scorers = evalCase.scorers ?? config.scorers;
|
|
28
|
-
const scores = {};
|
|
29
|
-
// Scores are keyed by name, so two scorers sharing one collapse:
|
|
30
|
-
// the mean's denominator becomes the count of distinct NAMES and
|
|
31
|
-
// the surviving score is whichever ran last. Two
|
|
32
|
-
// `containsScorer(...)` instances are both called 'contains', so
|
|
33
|
-
// this is easy to hit by accident and silently halves the
|
|
34
|
-
// evidence. Ambiguous results are worse than a loud failure.
|
|
35
|
-
const seen = new Set();
|
|
36
|
-
for (const scorer of scorers) {
|
|
37
|
-
if (seen.has(scorer.name)) {
|
|
38
|
-
throw new Error(`Duplicate scorer name "${scorer.name}" for case "${evalCase.name}". Scores are keyed by name, so the second would overwrite the first and the case mean would be computed over the wrong denominator. Give each scorer a distinct name.`);
|
|
39
|
-
}
|
|
40
|
-
seen.add(scorer.name);
|
|
41
|
-
scores[scorer.name] = await safeScore(scorer, run, evalCase);
|
|
42
|
-
}
|
|
43
|
-
// Only scores that were actually produced count — an unavailable
|
|
44
|
-
// scorer leaves the denominator alone rather than dragging the
|
|
45
|
-
// mean toward zero with a measurement that never happened.
|
|
46
|
-
const values = Object.values(scores)
|
|
47
|
-
.filter((s) => s.unavailable !== true)
|
|
48
|
-
.map((s) => s.score);
|
|
49
|
-
const mean = values.length === 0 ? 0 : values.reduce((a, b) => a + b, 0) / values.length;
|
|
50
|
-
// A gate miss fails the case whatever the mean says. Averaging a
|
|
51
|
-
// hard check together with a fuzzy one lets three good scores
|
|
52
|
-
// carry a zero: trajectory 0 + completion 1 + contains 1 +
|
|
53
|
-
// judge 1 averages to 0.75 and reports passed at a threshold of
|
|
54
|
-
// 0.75 — the exact regression the harness exists to catch,
|
|
55
|
-
// reported green. An UNAVAILABLE gate does not fail the case;
|
|
56
|
-
// it did not judge the run at all, which is the inconclusive
|
|
57
|
-
// path, not a failure.
|
|
58
|
-
const failedGates = scorers
|
|
59
|
-
.filter((s) => s.severity === 'gate')
|
|
60
|
-
.filter((s) => {
|
|
61
|
-
const score = scores[s.name];
|
|
62
|
-
if (!score || score.unavailable === true)
|
|
63
|
-
return false;
|
|
64
|
-
return score.score < (s.threshold ?? threshold);
|
|
65
|
-
})
|
|
66
|
-
.map((s) => s.name);
|
|
67
|
-
const status = Object.keys(scores).length > 0 && values.length === 0
|
|
68
|
-
? 'inconclusive'
|
|
69
|
-
: failedGates.length > 0
|
|
70
|
-
? 'failed'
|
|
71
|
-
: mean >= threshold
|
|
72
|
-
? 'passed'
|
|
73
|
-
: 'failed';
|
|
74
|
-
const result = {
|
|
75
|
-
case: evalCase.name,
|
|
76
|
-
run,
|
|
77
|
-
scores,
|
|
78
|
-
mean,
|
|
79
|
-
status,
|
|
80
|
-
passed: status === 'passed',
|
|
81
|
-
// Named, not just counted: "failed" with a mean of 0.75 sends
|
|
82
|
-
// somebody to read four scores and guess which one mattered.
|
|
83
|
-
...(failedGates.length > 0 ? { failedGates } : {}),
|
|
84
|
-
};
|
|
28
|
+
const result = await evaluateCase(config, evalCase, threshold, timeoutMs);
|
|
85
29
|
results[index] = result;
|
|
86
30
|
config.onCaseFinish?.(result);
|
|
87
31
|
}
|
|
@@ -107,36 +51,124 @@ export async function runExperiment(config) {
|
|
|
107
51
|
durationMs: Date.now() - startedAt,
|
|
108
52
|
};
|
|
109
53
|
}
|
|
54
|
+
function validateCaseTimeoutMs(value) {
|
|
55
|
+
if (value === undefined)
|
|
56
|
+
return undefined;
|
|
57
|
+
if (!Number.isInteger(value) || value < 1 || value > MAX_CASE_TIMEOUT_MS) {
|
|
58
|
+
throw new RangeError(`timeoutMs must be an integer from 1 to ${MAX_CASE_TIMEOUT_MS}, or omitted; received ${String(value)}`);
|
|
59
|
+
}
|
|
60
|
+
return value;
|
|
61
|
+
}
|
|
62
|
+
/** One wall-clock budget shared by execution and every scorer for a case. */
|
|
63
|
+
function openCaseDeadline(timeoutMs) {
|
|
64
|
+
const controller = new AbortController();
|
|
65
|
+
let timer;
|
|
66
|
+
let expiry;
|
|
67
|
+
let onAbort;
|
|
68
|
+
if (timeoutMs !== undefined && timeoutMs > 0) {
|
|
69
|
+
expiry = new Promise((_resolve, reject) => {
|
|
70
|
+
onAbort = () => reject(controller.signal.reason);
|
|
71
|
+
controller.signal.addEventListener('abort', onAbort, { once: true });
|
|
72
|
+
});
|
|
73
|
+
// A case can settle just before its timer. Keep the deadline promise
|
|
74
|
+
// observed independently so a same-tick expiry cannot become an
|
|
75
|
+
// unhandled rejection while the scope is being disposed.
|
|
76
|
+
void expiry.catch(() => { });
|
|
77
|
+
timer = setTimeout(() => controller.abort(new Error(`case timed out after ${timeoutMs}ms`)), timeoutMs);
|
|
78
|
+
}
|
|
79
|
+
return {
|
|
80
|
+
signal: controller.signal,
|
|
81
|
+
async race(work) {
|
|
82
|
+
if (!expiry)
|
|
83
|
+
return await work;
|
|
84
|
+
// A non-cooperative run or scorer may reject after the deadline won.
|
|
85
|
+
// Observe that loser without making suite settlement depend on it.
|
|
86
|
+
void work.catch(() => { });
|
|
87
|
+
return await Promise.race([work, expiry]);
|
|
88
|
+
},
|
|
89
|
+
dispose() {
|
|
90
|
+
clearTimeout(timer);
|
|
91
|
+
if (onAbort)
|
|
92
|
+
controller.signal.removeEventListener('abort', onAbort);
|
|
93
|
+
},
|
|
94
|
+
};
|
|
95
|
+
}
|
|
96
|
+
async function evaluateCase(config, evalCase, threshold, timeoutMs) {
|
|
97
|
+
const deadline = openCaseDeadline(timeoutMs);
|
|
98
|
+
try {
|
|
99
|
+
const run = await executeCase(config, evalCase, deadline);
|
|
100
|
+
const scorers = evalCase.scorers ?? config.scorers;
|
|
101
|
+
const scores = {};
|
|
102
|
+
// Scores are keyed by name, so two scorers sharing one collapse:
|
|
103
|
+
// the mean's denominator becomes the count of distinct NAMES and
|
|
104
|
+
// the surviving score is whichever ran last. Two
|
|
105
|
+
// `containsScorer(...)` instances are both called 'contains', so
|
|
106
|
+
// this is easy to hit by accident and silently halves the
|
|
107
|
+
// evidence. Ambiguous results are worse than a loud failure.
|
|
108
|
+
const seen = new Set();
|
|
109
|
+
for (const scorer of scorers) {
|
|
110
|
+
if (seen.has(scorer.name)) {
|
|
111
|
+
throw new Error(`Duplicate scorer name "${scorer.name}" for case "${evalCase.name}". Scores are keyed by name, so the second would overwrite the first and the case mean would be computed over the wrong denominator. Give each scorer a distinct name.`);
|
|
112
|
+
}
|
|
113
|
+
seen.add(scorer.name);
|
|
114
|
+
scores[scorer.name] = await safeScore(scorer, run, evalCase, deadline);
|
|
115
|
+
}
|
|
116
|
+
// Only scores that were actually produced count — an unavailable
|
|
117
|
+
// scorer leaves the denominator alone rather than dragging the
|
|
118
|
+
// mean toward zero with a measurement that never happened.
|
|
119
|
+
const values = Object.values(scores)
|
|
120
|
+
.filter((s) => s.unavailable !== true)
|
|
121
|
+
.map((s) => s.score);
|
|
122
|
+
const mean = values.length === 0 ? 0 : values.reduce((a, b) => a + b, 0) / values.length;
|
|
123
|
+
// A gate miss fails the case whatever the mean says. Averaging a
|
|
124
|
+
// hard check together with a fuzzy one lets three good scores
|
|
125
|
+
// carry a zero: trajectory 0 + completion 1 + contains 1 +
|
|
126
|
+
// judge 1 averages to 0.75 and reports passed at a threshold of
|
|
127
|
+
// 0.75 — the exact regression the harness exists to catch,
|
|
128
|
+
// reported green. An UNAVAILABLE gate does not fail the case;
|
|
129
|
+
// it did not judge the run at all, which is the inconclusive
|
|
130
|
+
// path, not a failure.
|
|
131
|
+
const failedGates = scorers
|
|
132
|
+
.filter((s) => s.severity === 'gate')
|
|
133
|
+
.filter((s) => {
|
|
134
|
+
const score = scores[s.name];
|
|
135
|
+
if (!score || score.unavailable === true)
|
|
136
|
+
return false;
|
|
137
|
+
return score.score < (s.threshold ?? threshold);
|
|
138
|
+
})
|
|
139
|
+
.map((s) => s.name);
|
|
140
|
+
const status = Object.keys(scores).length > 0 && values.length === 0
|
|
141
|
+
? 'inconclusive'
|
|
142
|
+
: failedGates.length > 0
|
|
143
|
+
? 'failed'
|
|
144
|
+
: mean >= threshold
|
|
145
|
+
? 'passed'
|
|
146
|
+
: 'failed';
|
|
147
|
+
return {
|
|
148
|
+
case: evalCase.name,
|
|
149
|
+
run,
|
|
150
|
+
scores,
|
|
151
|
+
mean,
|
|
152
|
+
status,
|
|
153
|
+
passed: status === 'passed',
|
|
154
|
+
// Named, not just counted: "failed" with a mean of 0.75 sends
|
|
155
|
+
// somebody to read four scores and guess which one mattered.
|
|
156
|
+
...(failedGates.length > 0 ? { failedGates } : {}),
|
|
157
|
+
};
|
|
158
|
+
}
|
|
159
|
+
finally {
|
|
160
|
+
deadline.dispose();
|
|
161
|
+
}
|
|
162
|
+
}
|
|
110
163
|
/**
|
|
111
164
|
* A case that throws is a RESULT, not a crash. An eval suite whose first
|
|
112
165
|
* broken case aborts the run tells you nothing about the other forty.
|
|
113
166
|
*/
|
|
114
|
-
async function executeCase(config, evalCase) {
|
|
167
|
+
async function executeCase(config, evalCase, deadline) {
|
|
115
168
|
const startedAt = Date.now();
|
|
116
|
-
const controller = new AbortController();
|
|
117
|
-
const timeoutMs = config.timeoutMs;
|
|
118
|
-
// The signal is handed to `run` so a closure that drives `query()` can
|
|
119
|
-
// pass it through and actually stop working. A closure that ignores it
|
|
120
|
-
// is merely detached rather than stopped — the same bargain every other
|
|
121
|
-
// deadline in the SDK makes — but the SUITE is unblocked either way,
|
|
122
|
-
// which is the part that was missing.
|
|
123
|
-
const timer = timeoutMs !== undefined && timeoutMs > 0
|
|
124
|
-
? setTimeout(() => controller.abort(new Error(`case timed out after ${timeoutMs}ms`)), timeoutMs)
|
|
125
|
-
: undefined;
|
|
126
|
-
const deadline = timer === undefined
|
|
127
|
-
? undefined
|
|
128
|
-
: new Promise((_resolve, reject) => {
|
|
129
|
-
controller.signal.addEventListener('abort', () => reject(controller.signal.reason), {
|
|
130
|
-
once: true,
|
|
131
|
-
});
|
|
132
|
-
});
|
|
133
169
|
try {
|
|
134
|
-
const work = config.run(evalCase.input, evalCase,
|
|
135
|
-
|
|
136
|
-
// rejection after the race has already been decided.
|
|
137
|
-
if (deadline)
|
|
138
|
-
work.catch(() => { });
|
|
139
|
-
return await (deadline ? Promise.race([work, deadline]) : work);
|
|
170
|
+
const work = config.run(evalCase.input, evalCase, deadline.signal);
|
|
171
|
+
return await deadline.race(work);
|
|
140
172
|
}
|
|
141
173
|
catch (err) {
|
|
142
174
|
return {
|
|
@@ -152,15 +184,9 @@ async function executeCase(config, evalCase) {
|
|
|
152
184
|
error: err instanceof Error ? err.message : String(err),
|
|
153
185
|
};
|
|
154
186
|
}
|
|
155
|
-
finally {
|
|
156
|
-
// Only the timer needs clearing. Aborting here would fire a spurious
|
|
157
|
-
// signal at a `run` that has already settled — the deadline branch
|
|
158
|
-
// has aborted already, and every other path is done.
|
|
159
|
-
clearTimeout(timer);
|
|
160
|
-
}
|
|
161
187
|
}
|
|
162
188
|
/** A throwing scorer scores zero with the throw as its reason. */
|
|
163
|
-
async function safeScore(scorer, run, evalCase) {
|
|
189
|
+
async function safeScore(scorer, run, evalCase, deadline) {
|
|
164
190
|
// A run that THREW scores zero, whatever the scorer would have said.
|
|
165
191
|
// `executeCase` catches the failure and returns an empty run, and an
|
|
166
192
|
// empty run walks straight into every scorer's happy path:
|
|
@@ -172,7 +198,11 @@ async function safeScore(scorer, run, evalCase) {
|
|
|
172
198
|
return { score: 0, reason: `run failed: ${run.error}`, details: { error: run.error } };
|
|
173
199
|
}
|
|
174
200
|
try {
|
|
175
|
-
|
|
201
|
+
deadline.signal.throwIfAborted();
|
|
202
|
+
// Defer invocation into the promise so a synchronous scorer throw is
|
|
203
|
+
// raced and observed by the same path as an asynchronous rejection.
|
|
204
|
+
const work = Promise.resolve().then(() => scorer.score(run, evalCase, deadline.signal));
|
|
205
|
+
return await deadline.race(work);
|
|
176
206
|
}
|
|
177
207
|
catch (err) {
|
|
178
208
|
// UNAVAILABLE, not zero. A scorer that threw did not judge the run
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"experiment.js","sourceRoot":"","sources":["../../src/eval/experiment.ts"],"names":[],"mappings":"AAUA,OAAO,EAAE,mBAAmB,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAA;AAmCrE;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,KAAK,UAAU,aAAa,CAClC,MAAgC;IAEhC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;IAC5B,MAAM,SAAS,GAAG,MAAM,CAAC,aAAa,IAAI,CAAC,CAAA;IAC3C,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,WAAW,IAAI,CAAC,CAAC,CAAA;IAExD,MAAM,OAAO,GAAiB,IAAI,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;IAC5D,IAAI,MAAM,GAAG,CAAC,CAAA;IAEd,MAAM,MAAM,GAAG,KAAK,IAAmB,EAAE;QACxC,SAAS,CAAC;YACT,MAAM,KAAK,GAAG,MAAM,EAAE,CAAA;YACtB,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;YACpC,IAAI,CAAC,QAAQ;gBAAE,OAAM;YAErB,MAAM,GAAG,GAAG,MAAM,WAAW,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAA;YAC/C,MAAM,OAAO,GAAG,QAAQ,CAAC,OAAO,IAAI,MAAM,CAAC,OAAO,CAAA;YAClD,MAAM,MAAM,GAA0B,EAAE,CAAA;YAExC,iEAAiE;YACjE,iEAAiE;YACjE,iDAAiD;YACjD,iEAAiE;YACjE,0DAA0D;YAC1D,6DAA6D;YAC7D,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;YAC9B,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;gBAC9B,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC3B,MAAM,IAAI,KAAK,CACd,0BAA0B,MAAM,CAAC,IAAI,eAAe,QAAQ,CAAC,IAAI,wKAAwK,CACzO,CAAA;gBACF,CAAC;gBACD,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;gBACrB,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,MAAM,SAAS,CAAC,MAAM,EAAE,GAAG,EAAE,QAAQ,CAAC,CAAA;YAC7D,CAAC;YAED,iEAAiE;YACjE,+DAA+D;YAC/D,2DAA2D;YAC3D,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC;iBAClC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,KAAK,IAAI,CAAC;iBACrC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAA;YACrB,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAA;YAExF,iEAAiE;YACjE,8DAA8D;YAC9D,2DAA2D;YAC3D,gEAAgE;YAChE,2DAA2D;YAC3D,8DAA8D;YAC9D,6DAA6D;YAC7D,uBAAuB;YACvB,MAAM,WAAW,GAAG,OAAO;iBACzB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,KAAK,MAAM,CAAC;iBACpC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE;gBACb,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;gBAC5B,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,WAAW,KAAK,IAAI;oBAAE,OAAO,KAAK,CAAA;gBACtD,OAAO,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,SAAS,IAAI,SAAS,CAAC,CAAA;YAChD,CAAC,CAAC;iBACD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;YAEpB,MAAM,MAAM,GACX,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;gBACpD,CAAC,CAAC,cAAc;gBAChB,CAAC,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC;oBACvB,CAAC,CAAC,QAAQ;oBACV,CAAC,CAAC,IAAI,IAAI,SAAS;wBAClB,CAAC,CAAC,QAAQ;wBACV,CAAC,CAAC,QAAQ,CAAA;YACd,MAAM,MAAM,GAAe;gBAC1B,IAAI,EAAE,QAAQ,CAAC,IAAI;gBACnB,GAAG;gBACH,MAAM;gBACN,IAAI;gBACJ,MAAM;gBACN,MAAM,EAAE,MAAM,KAAK,QAAQ;gBAC3B,8DAA8D;gBAC9D,6DAA6D;gBAC7D,GAAG,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;aAClD,CAAA;YACD,OAAO,CAAC,KAAK,CAAC,GAAG,MAAM,CAAA;YACvB,MAAM,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC,CAAA;IAED,MAAM,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAA;IAE7F,MAAM,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;IACvC,wEAAwE;IACxE,8DAA8D;IAC9D,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAA;IACjE,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAA;IAEjG,OAAO;QACN,IAAI,EAAE,MAAM,CAAC,IAAI;QACjB,KAAK,EAAE,OAAO;QACd,IAAI;QACJ,kEAAkE;QAClE,oEAAoE;QACpE,4DAA4D;QAC5D,WAAW,EAAE,aAAa,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QACrD,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,MAAM;QAC3D,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,MAAM;QAC3D,YAAY,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAC,MAAM;QACvE,QAAQ,EAAE,YAAY,CAAC,OAAO,CAAC;QAC/B,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS;KAClC,CAAA;AACF,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,WAAW,CACzB,MAAgC,EAChC,QAA0B;IAE1B,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;IAC5B,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAA;IACxC,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,CAAA;IAClC,uEAAuE;IACvE,uEAAuE;IACvE,wEAAwE;IACxE,qEAAqE;IACrE,sCAAsC;IACtC,MAAM,KAAK,GACV,SAAS,KAAK,SAAS,IAAI,SAAS,GAAG,CAAC;QACvC,CAAC,CAAC,UAAU,CACV,GAAG,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,wBAAwB,SAAS,IAAI,CAAC,CAAC,EACxE,SAAS,CACT;QACF,CAAC,CAAC,SAAS,CAAA;IAEb,MAAM,QAAQ,GACb,KAAK,KAAK,SAAS;QAClB,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,IAAI,OAAO,CAAQ,CAAC,QAAQ,EAAE,MAAM,EAAE,EAAE;YACxC,UAAU,CAAC,MAAM,CAAC,gBAAgB,CAAC,OAAO,EAAE,GAAG,EAAE,CAAC,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE;gBACnF,IAAI,EAAE,IAAI;aACV,CAAC,CAAA;QACH,CAAC,CAAC,CAAA;IAEL,IAAI,CAAC;QACJ,MAAM,IAAI,GAAG,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,EAAE,QAAQ,EAAE,UAAU,CAAC,MAAM,CAAC,CAAA;QACpE,8DAA8D;QAC9D,qDAAqD;QACrD,IAAI,QAAQ;YAAE,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;QAClC,OAAO,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;IAChE,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACd,OAAO;YACN,MAAM,EAAE,IAAI;YACZ,KAAK,EAAE,EAAE;YACT,SAAS,EAAE,EAAE;YACb,WAAW,EAAE,CAAC;YACd,YAAY,EAAE,CAAC;YACf,4DAA4D;YAC5D,6DAA6D;YAC7D,sBAAsB;YACtB,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS;YAClC,KAAK,EAAE,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC;SACvD,CAAA;IACF,CAAC;YAAS,CAAC;QACV,qEAAqE;QACrE,mEAAmE;QACnE,qDAAqD;QACrD,YAAY,CAAC,KAAK,CAAC,CAAA;IACpB,CAAC;AACF,CAAC;AAED,kEAAkE;AAClE,KAAK,UAAU,SAAS,CAAC,MAAc,EAAE,GAAY,EAAE,QAAkB;IACxE,qEAAqE;IACrE,qEAAqE;IACrE,2DAA2D;IAC3D,uEAAuE;IACvE,wEAAwE;IACxE,yEAAyE;IACzE,oDAAoD;IACpD,IAAI,GAAG,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,MAAM,EAAE,eAAe,GAAG,CAAC,KAAK,EAAE,EAAE,OAAO,EAAE,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,EAAE,EAAE,CAAA;IACvF,CAAC;IAED,IAAI,CAAC;QACJ,OAAO,MAAM,MAAM,CAAC,KAAK,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAA;IACzC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACd,mEAAmE;QACnE,6DAA6D;QAC7D,kEAAkE;QAClE,qEAAqE;QACrE,kEAAkE;QAClE,yCAAyC;QACzC,OAAO;YACN,KAAK,EAAE,CAAC;YACR,WAAW,EAAE,IAAI;YACjB,MAAM,EAAE,WAAW,MAAM,CAAC,IAAI,+BAC7B,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAChD,EAAE;SACF,CAAA;IACF,CAAC;AACF,CAAC;AAED,SAAS,YAAY,CAAC,OAA8B;IACnD,MAAM,IAAI,GAAG,IAAI,GAAG,EAA4C,CAAA;IAChE,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3D,IAAI,KAAK,CAAC,WAAW,KAAK,IAAI;gBAAE,SAAQ;YACxC,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,CAAA;YACtD,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,KAAK,CAAA;YAC1B,KAAK,CAAC,KAAK,EAAE,CAAA;YACb,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAA;QACtB,CAAC;IACF,CAAC;IACD,MAAM,GAAG,GAA2B,EAAE,CAAA;IACtC,oEAAoE;IACpE,sEAAsE;IACtE,qBAAqB;IACrB,KAAK,MAAM,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC;QAC7C,IAAI,KAAK,GAAG,CAAC;YAAE,GAAG,CAAC,IAAI,CAAC,GAAG,KAAK,GAAG,KAAK,CAAA;IACzC,CAAC;IACD,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;;;GAKG;AACH;;;;;;;;;;;;GAYG;AACH,SAAS,kBAAkB,CAAC,MAAwB;IACnD,OAAO,aAAa,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAA;AACjG,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,MAAwB;IACpD,MAAM,KAAK,GAAa;QACvB,GAAG,MAAM,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,iBAAiB,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,QAAQ,MAAM,CAAC,UAAU,IAAI;QAC3H,qEAAqE;QACrE,6DAA6D;QAC7D,qEAAqE;QACrE,gEAAgE;QAChE,mEAAmE;QACnE,qCAAqC;QACrC,KAAK,mBAAmB,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,WAAW,IAAI,kBAAkB,CAAC,MAAM,CAAC,CAAC,EAAE;QACzF,EAAE;KACF,CAAA;IAED,sEAAsE;IACtE,oEAAoE;IACpE,+BAA+B;IAC/B,IAAI,MAAM,CAAC,YAAY,GAAG,CAAC,EAAE,CAAC;QAC7B,KAAK,CAAC,IAAI,CACT,KAAK,MAAM,CAAC,YAAY,QAAQ,MAAM,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,yDAAyD,EAC7H,EAAE,CACF,CAAA;IACF,CAAC;IAED,KAAK,MAAM,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5D,KAAK,CAAC,IAAI,CAAC,KAAK,IAAI,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;IAC5C,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAA;IAClE,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,KAAK,CAAC,IAAI,CAAC,EAAE,EAAE,WAAW,CAAC,CAAA;QAC3B,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAChC,KAAK,CAAC,IAAI,CAAC,OAAO,OAAO,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAA;YAC9D,IAAI,OAAO,CAAC,WAAW,IAAI,OAAO,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC3D,8DAA8D;gBAC9D,uDAAuD;gBACvD,+BAA+B;gBAC/B,KAAK,CAAC,IAAI,CAAC,sBAAsB,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YACnE,CAAC;YACD,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC5D,IAAI,KAAK,CAAC,WAAW,KAAK,IAAI,EAAE,CAAC;oBAChC,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,kBAAkB,KAAK,CAAC,MAAM,EAAE,CAAC,CAAA;gBAC1D,CAAC;qBAAM,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;oBAC5B,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,KAAK,KAAK,CAAC,MAAM,EAAE,CAAC,CAAA;gBAC7C,CAAC;YACF,CAAC;YACD,IAAI,OAAO,CAAC,GAAG,CAAC,KAAK;gBAAE,KAAK,CAAC,IAAI,CAAC,gBAAgB,OAAO,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC,CAAA;QACvE,CAAC;IACF,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAA;IACxE,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,KAAK,CAAC,IAAI,CAAC,EAAE,EAAE,aAAa,CAAC,CAAA;QAC7B,KAAK,MAAM,IAAI,IAAI,QAAQ,EAAE,CAAC;YAC7B,KAAK,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,CAAA;YAC9B,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;gBACzD,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,KAAK,KAAK,CAAC,MAAM,EAAE,CAAC,CAAA;YAC7C,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AACxB,CAAC"}
|
|
1
|
+
{"version":3,"file":"experiment.js","sourceRoot":"","sources":["../../src/eval/experiment.ts"],"names":[],"mappings":"AAUA,OAAO,EAAE,mBAAmB,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAA;AAErE,MAAM,mBAAmB,GAAG,aAAa,CAAA;AAmCzC;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,KAAK,UAAU,aAAa,CAClC,MAAgC;IAEhC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;IAC5B,MAAM,SAAS,GAAG,MAAM,CAAC,aAAa,IAAI,CAAC,CAAA;IAC3C,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,WAAW,IAAI,CAAC,CAAC,CAAA;IACxD,MAAM,SAAS,GAAG,qBAAqB,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;IAEzD,MAAM,OAAO,GAAiB,IAAI,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;IAC5D,IAAI,MAAM,GAAG,CAAC,CAAA;IAEd,MAAM,MAAM,GAAG,KAAK,IAAmB,EAAE;QACxC,SAAS,CAAC;YACT,MAAM,KAAK,GAAG,MAAM,EAAE,CAAA;YACtB,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;YACpC,IAAI,CAAC,QAAQ;gBAAE,OAAM;YAErB,MAAM,MAAM,GAAG,MAAM,YAAY,CAAC,MAAM,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,CAAC,CAAA;YACzE,OAAO,CAAC,KAAK,CAAC,GAAG,MAAM,CAAA;YACvB,MAAM,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC,CAAA;IAED,MAAM,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC,CAAA;IAE7F,MAAM,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;IACvC,wEAAwE;IACxE,8DAA8D;IAC9D,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAA;IACjE,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAA;IAEjG,OAAO;QACN,IAAI,EAAE,MAAM,CAAC,IAAI;QACjB,KAAK,EAAE,OAAO;QACd,IAAI;QACJ,kEAAkE;QAClE,oEAAoE;QACpE,4DAA4D;QAC5D,WAAW,EAAE,aAAa,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QACrD,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,MAAM;QAC3D,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,MAAM;QAC3D,YAAY,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAC,MAAM;QACvE,QAAQ,EAAE,YAAY,CAAC,OAAO,CAAC;QAC/B,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS;KAClC,CAAA;AACF,CAAC;AAED,SAAS,qBAAqB,CAAC,KAAyB;IACvD,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO,SAAS,CAAA;IACzC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,KAAK,GAAG,CAAC,IAAI,KAAK,GAAG,mBAAmB,EAAE,CAAC;QAC1E,MAAM,IAAI,UAAU,CACnB,0CAA0C,mBAAmB,0BAA0B,MAAM,CAAC,KAAK,CAAC,EAAE,CACtG,CAAA;IACF,CAAC;IACD,OAAO,KAAK,CAAA;AACb,CAAC;AAQD,6EAA6E;AAC7E,SAAS,gBAAgB,CAAC,SAA6B;IACtD,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAA;IACxC,IAAI,KAAgD,CAAA;IACpD,IAAI,MAAkC,CAAA;IACtC,IAAI,OAAiC,CAAA;IAErC,IAAI,SAAS,KAAK,SAAS,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;QAC9C,MAAM,GAAG,IAAI,OAAO,CAAQ,CAAC,QAAQ,EAAE,MAAM,EAAE,EAAE;YAChD,OAAO,GAAG,GAAG,EAAE,CAAC,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;YAChD,UAAU,CAAC,MAAM,CAAC,gBAAgB,CAAC,OAAO,EAAE,OAAO,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAA;QACrE,CAAC,CAAC,CAAA;QACF,qEAAqE;QACrE,gEAAgE;QAChE,yDAAyD;QACzD,KAAK,MAAM,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;QAC3B,KAAK,GAAG,UAAU,CACjB,GAAG,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,wBAAwB,SAAS,IAAI,CAAC,CAAC,EACxE,SAAS,CACT,CAAA;IACF,CAAC;IAED,OAAO;QACN,MAAM,EAAE,UAAU,CAAC,MAAM;QACzB,KAAK,CAAC,IAAI,CAAI,IAAgB;YAC7B,IAAI,CAAC,MAAM;gBAAE,OAAO,MAAM,IAAI,CAAA;YAC9B,qEAAqE;YACrE,mEAAmE;YACnE,KAAK,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;YACzB,OAAO,MAAM,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC,CAAA;QAC1C,CAAC;QACD,OAAO;YACN,YAAY,CAAC,KAAK,CAAC,CAAA;YACnB,IAAI,OAAO;gBAAE,UAAU,CAAC,MAAM,CAAC,mBAAmB,CAAC,OAAO,EAAE,OAAO,CAAC,CAAA;QACrE,CAAC;KACD,CAAA;AACF,CAAC;AAED,KAAK,UAAU,YAAY,CAC1B,MAAgC,EAChC,QAA0B,EAC1B,SAAiB,EACjB,SAA6B;IAE7B,MAAM,QAAQ,GAAG,gBAAgB,CAAC,SAAS,CAAC,CAAA;IAC5C,IAAI,CAAC;QACJ,MAAM,GAAG,GAAG,MAAM,WAAW,CAAC,MAAM,EAAE,QAAQ,EAAE,QAAQ,CAAC,CAAA;QACzD,MAAM,OAAO,GAAG,QAAQ,CAAC,OAAO,IAAI,MAAM,CAAC,OAAO,CAAA;QAClD,MAAM,MAAM,GAA0B,EAAE,CAAA;QAExC,iEAAiE;QACjE,iEAAiE;QACjE,iDAAiD;QACjD,iEAAiE;QACjE,0DAA0D;QAC1D,6DAA6D;QAC7D,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;QAC9B,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC9B,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC;gBAC3B,MAAM,IAAI,KAAK,CACd,0BAA0B,MAAM,CAAC,IAAI,eAAe,QAAQ,CAAC,IAAI,wKAAwK,CACzO,CAAA;YACF,CAAC;YACD,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;YACrB,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,MAAM,SAAS,CAAC,MAAM,EAAE,GAAG,EAAE,QAAQ,EAAE,QAAQ,CAAC,CAAA;QACvE,CAAC;QAED,iEAAiE;QACjE,+DAA+D;QAC/D,2DAA2D;QAC3D,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC;aAClC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,KAAK,IAAI,CAAC;aACrC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAA;QACrB,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAA;QAExF,iEAAiE;QACjE,8DAA8D;QAC9D,2DAA2D;QAC3D,gEAAgE;QAChE,2DAA2D;QAC3D,8DAA8D;QAC9D,6DAA6D;QAC7D,uBAAuB;QACvB,MAAM,WAAW,GAAG,OAAO;aACzB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,KAAK,MAAM,CAAC;aACpC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE;YACb,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;YAC5B,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,WAAW,KAAK,IAAI;gBAAE,OAAO,KAAK,CAAA;YACtD,OAAO,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,SAAS,IAAI,SAAS,CAAC,CAAA;QAChD,CAAC,CAAC;aACD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QAEpB,MAAM,MAAM,GACX,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YACpD,CAAC,CAAC,cAAc;YAChB,CAAC,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC;gBACvB,CAAC,CAAC,QAAQ;gBACV,CAAC,CAAC,IAAI,IAAI,SAAS;oBAClB,CAAC,CAAC,QAAQ;oBACV,CAAC,CAAC,QAAQ,CAAA;QACd,OAAO;YACN,IAAI,EAAE,QAAQ,CAAC,IAAI;YACnB,GAAG;YACH,MAAM;YACN,IAAI;YACJ,MAAM;YACN,MAAM,EAAE,MAAM,KAAK,QAAQ;YAC3B,8DAA8D;YAC9D,6DAA6D;YAC7D,GAAG,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SAClD,CAAA;IACF,CAAC;YAAS,CAAC;QACV,QAAQ,CAAC,OAAO,EAAE,CAAA;IACnB,CAAC;AACF,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,WAAW,CACzB,MAAgC,EAChC,QAA0B,EAC1B,QAAsB;IAEtB,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;IAC5B,IAAI,CAAC;QACJ,MAAM,IAAI,GAAG,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,EAAE,QAAQ,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAA;QAClE,OAAO,MAAM,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IACjC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACd,OAAO;YACN,MAAM,EAAE,IAAI;YACZ,KAAK,EAAE,EAAE;YACT,SAAS,EAAE,EAAE;YACb,WAAW,EAAE,CAAC;YACd,YAAY,EAAE,CAAC;YACf,4DAA4D;YAC5D,6DAA6D;YAC7D,sBAAsB;YACtB,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS;YAClC,KAAK,EAAE,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC;SACvD,CAAA;IACF,CAAC;AACF,CAAC;AAED,kEAAkE;AAClE,KAAK,UAAU,SAAS,CACvB,MAAc,EACd,GAAY,EACZ,QAAkB,EAClB,QAAsB;IAEtB,qEAAqE;IACrE,qEAAqE;IACrE,2DAA2D;IAC3D,uEAAuE;IACvE,wEAAwE;IACxE,yEAAyE;IACzE,oDAAoD;IACpD,IAAI,GAAG,CAAC,KAAK,KAAK,SAAS,EAAE,CAAC;QAC7B,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,MAAM,EAAE,eAAe,GAAG,CAAC,KAAK,EAAE,EAAE,OAAO,EAAE,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,EAAE,EAAE,CAAA;IACvF,CAAC;IAED,IAAI,CAAC;QACJ,QAAQ,CAAC,MAAM,CAAC,cAAc,EAAE,CAAA;QAChC,qEAAqE;QACrE,oEAAoE;QACpE,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,EAAE,QAAQ,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAA;QACvF,OAAO,MAAM,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IACjC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACd,mEAAmE;QACnE,6DAA6D;QAC7D,kEAAkE;QAClE,qEAAqE;QACrE,kEAAkE;QAClE,yCAAyC;QACzC,OAAO;YACN,KAAK,EAAE,CAAC;YACR,WAAW,EAAE,IAAI;YACjB,MAAM,EAAE,WAAW,MAAM,CAAC,IAAI,+BAC7B,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAChD,EAAE;SACF,CAAA;IACF,CAAC;AACF,CAAC;AAED,SAAS,YAAY,CAAC,OAA8B;IACnD,MAAM,IAAI,GAAG,IAAI,GAAG,EAA4C,CAAA;IAChE,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3D,IAAI,KAAK,CAAC,WAAW,KAAK,IAAI;gBAAE,SAAQ;YACxC,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,CAAA;YACtD,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,KAAK,CAAA;YAC1B,KAAK,CAAC,KAAK,EAAE,CAAA;YACb,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAA;QACtB,CAAC;IACF,CAAC;IACD,MAAM,GAAG,GAA2B,EAAE,CAAA;IACtC,oEAAoE;IACpE,sEAAsE;IACtE,qBAAqB;IACrB,KAAK,MAAM,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC;QAC7C,IAAI,KAAK,GAAG,CAAC;YAAE,GAAG,CAAC,IAAI,CAAC,GAAG,KAAK,GAAG,KAAK,CAAA;IACzC,CAAC;IACD,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;;;GAKG;AACH;;;;;;;;;;;;GAYG;AACH,SAAS,kBAAkB,CAAC,MAAwB;IACnD,OAAO,aAAa,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAA;AACjG,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,MAAwB;IACpD,MAAM,KAAK,GAAa;QACvB,GAAG,MAAM,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,iBAAiB,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,QAAQ,MAAM,CAAC,UAAU,IAAI;QAC3H,qEAAqE;QACrE,6DAA6D;QAC7D,qEAAqE;QACrE,gEAAgE;QAChE,mEAAmE;QACnE,qCAAqC;QACrC,KAAK,mBAAmB,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,WAAW,IAAI,kBAAkB,CAAC,MAAM,CAAC,CAAC,EAAE;QACzF,EAAE;KACF,CAAA;IAED,sEAAsE;IACtE,oEAAoE;IACpE,+BAA+B;IAC/B,IAAI,MAAM,CAAC,YAAY,GAAG,CAAC,EAAE,CAAC;QAC7B,KAAK,CAAC,IAAI,CACT,KAAK,MAAM,CAAC,YAAY,QAAQ,MAAM,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,yDAAyD,EAC7H,EAAE,CACF,CAAA;IACF,CAAC;IAED,KAAK,MAAM,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5D,KAAK,CAAC,IAAI,CAAC,KAAK,IAAI,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;IAC5C,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAA;IAClE,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,KAAK,CAAC,IAAI,CAAC,EAAE,EAAE,WAAW,CAAC,CAAA;QAC3B,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAChC,KAAK,CAAC,IAAI,CAAC,OAAO,OAAO,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAA;YAC9D,IAAI,OAAO,CAAC,WAAW,IAAI,OAAO,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC3D,8DAA8D;gBAC9D,uDAAuD;gBACvD,+BAA+B;gBAC/B,KAAK,CAAC,IAAI,CAAC,sBAAsB,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YACnE,CAAC;YACD,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC5D,IAAI,KAAK,CAAC,WAAW,KAAK,IAAI,EAAE,CAAC;oBAChC,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,kBAAkB,KAAK,CAAC,MAAM,EAAE,CAAC,CAAA;gBAC1D,CAAC;qBAAM,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;oBAC5B,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,KAAK,KAAK,CAAC,MAAM,EAAE,CAAC,CAAA;gBAC7C,CAAC;YACF,CAAC;YACD,IAAI,OAAO,CAAC,GAAG,CAAC,KAAK;gBAAE,KAAK,CAAC,IAAI,CAAC,gBAAgB,OAAO,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC,CAAA;QACvE,CAAC;IACF,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,cAAc,CAAC,CAAA;IACxE,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,KAAK,CAAC,IAAI,CAAC,EAAE,EAAE,aAAa,CAAC,CAAA;QAC7B,KAAK,MAAM,IAAI,IAAI,QAAQ,EAAE,CAAC;YAC7B,KAAK,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,CAAA;YAC9B,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;gBACzD,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,KAAK,KAAK,CAAC,MAAM,EAAE,CAAC,CAAA;YAC7C,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AACxB,CAAC"}
|
package/dist/eval/judge.d.ts
CHANGED
|
@@ -58,6 +58,12 @@ export interface JudgeScorerConfig {
|
|
|
58
58
|
* truncation rather than the run.
|
|
59
59
|
*/
|
|
60
60
|
maxOutputChars?: number;
|
|
61
|
+
/**
|
|
62
|
+
* Milliseconds without a judge chunk before its provider stream is
|
|
63
|
+
* treated as stalled. Defaults to the SDK's finite provider-stream idle
|
|
64
|
+
* bound. Set `0` only for explicit unbounded compatibility.
|
|
65
|
+
*/
|
|
66
|
+
streamIdleTimeoutMs?: number;
|
|
61
67
|
}
|
|
62
68
|
export declare function judgeScorer(config: JudgeScorerConfig): Scorer;
|
|
63
69
|
//# sourceMappingURL=judge.d.ts.map
|
package/dist/eval/judge.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"judge.d.ts","sourceRoot":"","sources":["../../src/eval/judge.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"judge.d.ts","sourceRoot":"","sources":["../../src/eval/judge.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,gCAAgC,CAAA;AACjE,OAAO,KAAK,EAA4B,MAAM,EAAE,MAAM,YAAY,CAAA;AAElE;;;;;;;;;;;;;;GAcG;AACH,MAAM,WAAW,iBAAiB;IACjC,qEAAqE;IACrE,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,QAAQ,EAAE,WAAW,CAAA;IACrB,KAAK,EAAE,MAAM,CAAA;IACb;;;;;;;OAOG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;;;;;;;;;;OAWG;IACH,KAAK,CAAC,EAAE,MAAM,CAAA;IACd;;;;;;OAMG;IACH,iBAAiB,CAAC,EAAE,OAAO,CAAA;IAC3B;;;;;;OAMG;IACH,cAAc,CAAC,EAAE,MAAM,CAAA;IACvB;;;;OAIG;IACH,mBAAmB,CAAC,EAAE,MAAM,CAAA;CAC5B;AAwID,wBAAgB,WAAW,CAAC,MAAM,EAAE,iBAAiB,GAAG,MAAM,CA6C7D"}
|
package/dist/eval/judge.js
CHANGED
|
@@ -1,4 +1,5 @@
|
|
|
1
1
|
import { collectChatCompletion } from '../provider/collect-chat-completion.js';
|
|
2
|
+
import { resolveStreamIdleTimeoutMs, withStreamIdleTimeout } from '../provider/idle-timeout.js';
|
|
2
3
|
const DEFAULT_SCALE = 4;
|
|
3
4
|
const DEFAULT_MAX_OUTPUT_CHARS = 20_000;
|
|
4
5
|
/**
|
|
@@ -99,6 +100,9 @@ function buildPrompt(config, run, evalCase, scale) {
|
|
|
99
100
|
}
|
|
100
101
|
export function judgeScorer(config) {
|
|
101
102
|
const scale = config.scale ?? DEFAULT_SCALE;
|
|
103
|
+
const provider = withStreamIdleTimeout(config.provider, {
|
|
104
|
+
idleTimeoutMs: resolveStreamIdleTimeoutMs(config.streamIdleTimeoutMs),
|
|
105
|
+
});
|
|
102
106
|
if (!Number.isInteger(scale) || scale < 1) {
|
|
103
107
|
throw new Error(`judgeScorer: scale must be a positive integer, got ${scale}`);
|
|
104
108
|
}
|
|
@@ -107,14 +111,15 @@ export function judgeScorer(config) {
|
|
|
107
111
|
}
|
|
108
112
|
return {
|
|
109
113
|
name: config.name ?? 'judge',
|
|
110
|
-
async score(run, evalCase) {
|
|
111
|
-
const response = await collectChatCompletion(
|
|
114
|
+
async score(run, evalCase, signal) {
|
|
115
|
+
const response = await collectChatCompletion(provider.chatStream({
|
|
112
116
|
model: config.model,
|
|
113
117
|
messages: [{ role: 'user', content: buildPrompt(config, run, evalCase, scale) }],
|
|
114
118
|
// The same run must grade the same way twice, or a
|
|
115
119
|
// regression cannot be told from sampling noise.
|
|
116
120
|
temperature: 0,
|
|
117
121
|
maxTokens: 512,
|
|
122
|
+
...(signal ? { signal } : {}),
|
|
118
123
|
}));
|
|
119
124
|
const verdict = parseVerdict(response.message.content ?? '', scale);
|
|
120
125
|
return {
|
package/dist/eval/judge.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"judge.js","sourceRoot":"","sources":["../../src/eval/judge.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,qBAAqB,EAAE,MAAM,wCAAwC,CAAA;
|
|
1
|
+
{"version":3,"file":"judge.js","sourceRoot":"","sources":["../../src/eval/judge.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,qBAAqB,EAAE,MAAM,wCAAwC,CAAA;AAC9E,OAAO,EAAE,0BAA0B,EAAE,qBAAqB,EAAE,MAAM,6BAA6B,CAAA;AAsE/F,MAAM,aAAa,GAAG,CAAC,CAAA;AACvB,MAAM,wBAAwB,GAAG,MAAM,CAAA;AAOvC;;;;;;;;GAQG;AACH,SAAS,YAAY,CAAC,KAAa,EAAE,KAAa;IACjD,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;IAChC,IAAI,KAAK,KAAK,CAAC,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,kCAAkC,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;IAE1F,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,IAAI,QAAQ,GAAG,KAAK,CAAA;IACpB,IAAI,OAAO,GAAG,KAAK,CAAA;IACnB,IAAI,GAAG,GAAG,CAAC,CAAC,CAAA;IACZ,KAAK,IAAI,CAAC,GAAG,KAAK,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC3C,MAAM,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC,CAAA;QACnB,IAAI,OAAO,EAAE,CAAC;YACb,OAAO,GAAG,KAAK,CAAA;YACf,SAAQ;QACT,CAAC;QACD,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;YACjB,OAAO,GAAG,IAAI,CAAA;YACd,SAAQ;QACT,CAAC;QACD,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YAChB,QAAQ,GAAG,CAAC,QAAQ,CAAA;YACpB,SAAQ;QACT,CAAC;QACD,IAAI,QAAQ;YAAE,SAAQ;QACtB,IAAI,EAAE,KAAK,GAAG;YAAE,KAAK,EAAE,CAAA;aAClB,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACrB,KAAK,EAAE,CAAA;YACP,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACjB,GAAG,GAAG,CAAC,CAAA;gBACP,MAAK;YACN,CAAC;QACF,CAAC;IACF,CAAC;IACD,IAAI,GAAG,KAAK,CAAC,CAAC;QACb,MAAM,IAAI,KAAK,CAAC,+CAA+C,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;IAEtF,IAAI,MAAe,CAAA;IACnB,IAAI,CAAC;QACJ,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC,CAAA;IACjD,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACd,MAAM,IAAI,KAAK,CACd,mCAAmC,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,CACrF,CAAA;IACF,CAAC;IAED,MAAM,KAAK,GAAI,MAA8B,CAAC,KAAK,CAAA;IACnD,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QAC1D,MAAM,IAAI,KAAK,CAAC,oCAAoC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;IAC5F,CAAC;IACD,IAAI,KAAK,GAAG,CAAC,IAAI,KAAK,GAAG,KAAK,EAAE,CAAC;QAChC,qEAAqE;QACrE,qEAAqE;QACrE,2BAA2B;QAC3B,MAAM,IAAI,KAAK,CAAC,kBAAkB,KAAK,oBAAoB,KAAK,qBAAqB,CAAC,CAAA;IACvF,CAAC;IAED,MAAM,MAAM,GAAI,MAA+B,CAAC,MAAM,CAAA;IACtD,OAAO;QACN,KAAK;QACL,MAAM,EAAE,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,wBAAwB;KAC3F,CAAA;AACF,CAAC;AAED,SAAS,WAAW,CACnB,MAAyB,EACzB,GAAY,EACZ,QAAkB,EAClB,KAAa;IAEb,MAAM,KAAK,GAAG,MAAM,CAAC,cAAc,IAAI,wBAAwB,CAAA;IAC/D,MAAM,MAAM,GAAG,GAAG,CAAC,MAAM,IAAI,EAAE,CAAA;IAC/B,MAAM,SAAS,GAAG,MAAM,CAAC,MAAM,GAAG,KAAK,CAAA;IAEvC,MAAM,QAAQ,GAAa;QAC1B,sCAAsC;QACtC,EAAE;QACF,YAAY,MAAM,CAAC,MAAM,EAAE;QAC3B,EAAE;QACF,UAAU,OAAO,QAAQ,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE;KAChG,CAAA;IAED,IAAI,QAAQ,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;QACrC,QAAQ,CAAC,IAAI,CACZ,EAAE,EACF,mDACC,OAAO,QAAQ,CAAC,QAAQ,KAAK,QAAQ;YACpC,CAAC,CAAC,QAAQ,CAAC,QAAQ;YACnB,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,QAAQ,CACpC,EAAE,CACF,CAAA;IACF,CAAC;IAED,IAAI,MAAM,CAAC,iBAAiB,KAAK,IAAI,EAAE,CAAC;QACvC,QAAQ,CAAC,IAAI,CACZ,EAAE,EACF,4BAA4B,GAAG,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,CAC9F,CAAA;IACF,CAAC;IAED,QAAQ,CAAC,IAAI,CAAC,EAAE,EAAE,YAAY,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,CAAA;IAC5E,IAAI,SAAS,EAAE,CAAC;QACf,+DAA+D;QAC/D,qBAAqB;QACrB,QAAQ,CAAC,IAAI,CACZ,EAAE,EACF,gCAAgC,KAAK,wFAAwF,CAC7H,CAAA;IACF,CAAC;IAED,QAAQ,CAAC,IAAI,CACZ,EAAE,EACF,2DAA2D,KAAK,yCAAyC,EACzG,yCAAyC,KAAK,wDAAwD,CACtG,CAAA;IAED,OAAO,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AAC3B,CAAC;AAED,MAAM,UAAU,WAAW,CAAC,MAAyB;IACpD,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,aAAa,CAAA;IAC3C,MAAM,QAAQ,GAAG,qBAAqB,CAAC,MAAM,CAAC,QAAQ,EAAE;QACvD,aAAa,EAAE,0BAA0B,CAAC,MAAM,CAAC,mBAAmB,CAAC;KACrE,CAAC,CAAA;IACF,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;QAC3C,MAAM,IAAI,KAAK,CAAC,sDAAsD,KAAK,EAAE,CAAC,CAAA;IAC/E,CAAC;IACD,IAAI,MAAM,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvC,MAAM,IAAI,KAAK,CACd,qHAAqH,CACrH,CAAA;IACF,CAAC;IAED,OAAO;QACN,IAAI,EAAE,MAAM,CAAC,IAAI,IAAI,OAAO;QAC5B,KAAK,CAAC,KAAK,CAAC,GAAY,EAAE,QAAkB,EAAE,MAAoB;YACjE,MAAM,QAAQ,GAAG,MAAM,qBAAqB,CAC3C,QAAQ,CAAC,UAAU,CAAC;gBACnB,KAAK,EAAE,MAAM,CAAC,KAAK;gBACnB,QAAQ,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,WAAW,CAAC,MAAM,EAAE,GAAG,EAAE,QAAQ,EAAE,KAAK,CAAC,EAAE,CAAC;gBAChF,mDAAmD;gBACnD,iDAAiD;gBACjD,WAAW,EAAE,CAAC;gBACd,SAAS,EAAE,GAAG;gBACd,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;aAC7B,CAAC,CACF,CAAA;YAED,MAAM,OAAO,GAAG,YAAY,CAAC,QAAQ,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,EAAE,KAAK,CAAC,CAAA;YACnE,OAAO;gBACN,KAAK,EAAE,OAAO,CAAC,KAAK,GAAG,KAAK;gBAC5B,MAAM,EAAE,OAAO,CAAC,MAAM;gBACtB,OAAO,EAAE;oBACR,KAAK,EAAE,OAAO,CAAC,KAAK;oBACpB,KAAK;oBACL,sDAAsD;oBACtD,0DAA0D;oBAC1D,uDAAuD;oBACvD,YAAY;oBACZ,WAAW,EAAE,QAAQ,CAAC,KAAK,CAAC,WAAW;iBACvC;aACD,CAAA;QACF,CAAC;KACD,CAAA;AACF,CAAC"}
|
package/dist/eval/types.d.ts
CHANGED
|
@@ -93,7 +93,13 @@ export interface Score {
|
|
|
93
93
|
export type ScorerSeverity = 'gate' | 'soft';
|
|
94
94
|
export interface Scorer {
|
|
95
95
|
name: string;
|
|
96
|
-
|
|
96
|
+
/**
|
|
97
|
+
* Produce one measurement. `runExperiment` passes the case's shared
|
|
98
|
+
* deadline signal, which covers both execution and every scorer. A pure
|
|
99
|
+
* scorer may ignore it; a scorer that owns I/O should carry it to that
|
|
100
|
+
* transport so timed-out work is stopped rather than merely abandoned.
|
|
101
|
+
*/
|
|
102
|
+
score(run: EvalRun, evalCase: EvalCase, signal?: AbortSignal): Score | Promise<Score>;
|
|
97
103
|
/** See {@link ScorerSeverity}. Default `soft`. */
|
|
98
104
|
severity?: ScorerSeverity;
|
|
99
105
|
/**
|
package/dist/eval/types.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/eval/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAA;AACtD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAA;AAExD;;;;;GAKG;AACH,MAAM,WAAW,QAAQ,CAAC,MAAM,GAAG,OAAO;IACzC,IAAI,EAAE,MAAM,CAAA;IACZ,KAAK,EAAE,MAAM,CAAA;IACb,uEAAuE;IACvE,QAAQ,CAAC,EAAE,OAAO,CAAA;IAClB;;;OAGG;IACH,aAAa,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;IACjC,sEAAsE;IACtE,OAAO,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CAC3B;AAED,sDAAsD;AACtD,MAAM,WAAW,OAAO;IACvB,uDAAuD;IACvD,MAAM,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,4EAA4E;IAC5E,gBAAgB,CAAC,EAAE,OAAO,CAAA;IAC1B,KAAK,EAAE,SAAS,UAAU,EAAE,CAAA;IAC5B,wDAAwD;IACxD,SAAS,EAAE,SAAS,MAAM,EAAE,CAAA;IAC5B,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,WAAW,EAAE,MAAM,CAAA;IACnB,YAAY,EAAE,MAAM,CAAA;IACpB,UAAU,EAAE,MAAM,CAAA;IAClB,mDAAmD;IACnD,KAAK,CAAC,EAAE,MAAM,CAAA;CACd;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,KAAK;IACrB,iEAAiE;IACjE,KAAK,EAAE,MAAM,CAAA;IACb,MAAM,EAAE,MAAM,CAAA;IACd,2EAA2E;IAC3E,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IACjC;;;;;;;;;;;;;;;;OAgBG;IACH,WAAW,CAAC,EAAE,OAAO,CAAA;CACrB;AAED;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,MAAM,MAAM,cAAc,GAAG,MAAM,GAAG,MAAM,CAAA;AAE5C,MAAM,WAAW,MAAM;IACtB,IAAI,EAAE,MAAM,CAAA;IACZ,KAAK,CAAC,GAAG,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,GAAG,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAA;
|
|
1
|
+
{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/eval/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAA;AACtD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAA;AAExD;;;;;GAKG;AACH,MAAM,WAAW,QAAQ,CAAC,MAAM,GAAG,OAAO;IACzC,IAAI,EAAE,MAAM,CAAA;IACZ,KAAK,EAAE,MAAM,CAAA;IACb,uEAAuE;IACvE,QAAQ,CAAC,EAAE,OAAO,CAAA;IAClB;;;OAGG;IACH,aAAa,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;IACjC,sEAAsE;IACtE,OAAO,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CAC3B;AAED,sDAAsD;AACtD,MAAM,WAAW,OAAO;IACvB,uDAAuD;IACvD,MAAM,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,4EAA4E;IAC5E,gBAAgB,CAAC,EAAE,OAAO,CAAA;IAC1B,KAAK,EAAE,SAAS,UAAU,EAAE,CAAA;IAC5B,wDAAwD;IACxD,SAAS,EAAE,SAAS,MAAM,EAAE,CAAA;IAC5B,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,WAAW,EAAE,MAAM,CAAA;IACnB,YAAY,EAAE,MAAM,CAAA;IACpB,UAAU,EAAE,MAAM,CAAA;IAClB,mDAAmD;IACnD,KAAK,CAAC,EAAE,MAAM,CAAA;CACd;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,KAAK;IACrB,iEAAiE;IACjE,KAAK,EAAE,MAAM,CAAA;IACb,MAAM,EAAE,MAAM,CAAA;IACd,2EAA2E;IAC3E,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IACjC;;;;;;;;;;;;;;;;OAgBG;IACH,WAAW,CAAC,EAAE,OAAO,CAAA;CACrB;AAED;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,MAAM,MAAM,cAAc,GAAG,MAAM,GAAG,MAAM,CAAA;AAE5C,MAAM,WAAW,MAAM;IACtB,IAAI,EAAE,MAAM,CAAA;IACZ;;;;;OAKG;IACH,KAAK,CAAC,GAAG,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,CAAC,EAAE,WAAW,GAAG,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAA;IACrF,kDAAkD;IAClD,QAAQ,CAAC,EAAE,cAAc,CAAA;IACzB;;;;;;;OAOG;IACH,SAAS,CAAC,EAAE,MAAM,CAAA;CAClB;AAED;;;;;GAKG;AACH,MAAM,MAAM,UAAU,GAAG,QAAQ,GAAG,QAAQ,GAAG,cAAc,CAAA;AAE7D,MAAM,WAAW,UAAU;IAC1B,IAAI,EAAE,MAAM,CAAA;IACZ,GAAG,EAAE,OAAO,CAAA;IACZ,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,KAAK,CAAC,CAAA;IAC7B,iEAAiE;IACjE,IAAI,EAAE,MAAM,CAAA;IACZ,MAAM,EAAE,UAAU,CAAA;IAClB,wCAAwC;IACxC,MAAM,EAAE,OAAO,CAAA;IACf;;;;;;OAMG;IACH,WAAW,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CAC/B;AAED,MAAM,WAAW,gBAAgB;IAChC,IAAI,EAAE,MAAM,CAAA;IACZ,KAAK,EAAE,SAAS,UAAU,EAAE,CAAA;IAC5B,qDAAqD;IACrD,IAAI,EAAE,MAAM,CAAA;IACZ;;;;;;;;;;;;;;OAcG;IACH,WAAW,CAAC,EAAE,gBAAgB,CAAA;IAC9B,MAAM,EAAE,MAAM,CAAA;IACd,MAAM,EAAE,MAAM,CAAA;IACd,uDAAuD;IACvD,YAAY,EAAE,MAAM,CAAA;IACpB;;;;OAIG;IACH,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAChC,UAAU,EAAE,MAAM,CAAA;CAClB"}
|
|
@@ -12,7 +12,7 @@ import type { SubSessionId } from '../../types/session/ids.js';
|
|
|
12
12
|
import type { SessionStore } from '../../types/session/store.js';
|
|
13
13
|
import type { WorkspaceRef } from '../../types/workspace/ref.js';
|
|
14
14
|
import { type Logger } from '../../utils/logger.js';
|
|
15
|
-
import type
|
|
15
|
+
import { type TopicManagerDependency } from '../topic/dependency.js';
|
|
16
16
|
/**
|
|
17
17
|
* Dependencies threaded into {@link AgentManager}. Phase 6 promoted the
|
|
18
18
|
* SubSession + Session + WorkspaceRef triple to mandatory spawn primitives —
|
|
@@ -32,19 +32,11 @@ import type { TopicManager } from '../topic/lifecycle.js';
|
|
|
32
32
|
* registry deny-by-default while matching pattern doc §7.1 (lazy workspace
|
|
33
33
|
* provisioning).
|
|
34
34
|
*/
|
|
35
|
-
|
|
35
|
+
interface AgentManagerBaseDeps {
|
|
36
36
|
readonly sessionStore: SessionStore;
|
|
37
37
|
readonly workspaceRegistry: WorkspaceBackendRegistry;
|
|
38
38
|
readonly summaryMaterializer: SessionSummaryMaterializer;
|
|
39
39
|
readonly capacity: CapacityValidator;
|
|
40
|
-
/**
|
|
41
|
-
* Gate session creation on the parent Topic being `'open'` via
|
|
42
|
-
* {@link TopicManager.requireOpen}. Added in Phase 2.6 to close the
|
|
43
|
-
* archive-gate gap flagged by the Phase 2.5 commit: without this,
|
|
44
|
-
* `TopicManager.archive` was best-effort because spawn could still
|
|
45
|
-
* attach a live session under an archived Thread.
|
|
46
|
-
*/
|
|
47
|
-
readonly threadManager: TopicManager;
|
|
48
40
|
/**
|
|
49
41
|
* A pre-built logger. No in-package caller threads a real one today —
|
|
50
42
|
* `packages/cli/src/integrations/subagents/runtime.ts` is the only
|
|
@@ -58,6 +50,15 @@ export interface AgentManagerDeps {
|
|
|
58
50
|
*/
|
|
59
51
|
readonly log?: Logger;
|
|
60
52
|
}
|
|
53
|
+
/**
|
|
54
|
+
* Dependencies for {@link AgentManager}.
|
|
55
|
+
*
|
|
56
|
+
* `topicManager` gates child-session creation on the parent Topic being open.
|
|
57
|
+
* The deprecated `threadManager` spelling remains accepted for one migration
|
|
58
|
+
* window through {@link TopicManagerDependency}; new callers use
|
|
59
|
+
* `topicManager`.
|
|
60
|
+
*/
|
|
61
|
+
export type AgentManagerDeps = AgentManagerBaseDeps & TopicManagerDependency;
|
|
61
62
|
interface ChildSpawnRecord {
|
|
62
63
|
subSessionId: SubSessionId;
|
|
63
64
|
childSessionId: SessionId;
|
|
@@ -87,6 +88,7 @@ export declare class AgentManager {
|
|
|
87
88
|
/** One provisioning at a time per parent — see {@link provisionSpawn}. */
|
|
88
89
|
private spawnLocks;
|
|
89
90
|
private deps;
|
|
91
|
+
private topicManager;
|
|
90
92
|
constructor(registry: AgentRegistry, config: Partial<AgentManagerConfig> | undefined, deps: AgentManagerDeps);
|
|
91
93
|
sendMessage(options: SendMessageOptions, context: AgentTaskContext, listener?: RunEventListener): Promise<AgentTask>;
|
|
92
94
|
cancel(taskId: TaskId, cause?: CancelCause): void;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"lifecycle.d.ts","sourceRoot":"","sources":["../../../src/manager/agent/lifecycle.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qCAAqC,CAAA;AACxE,OAAO,EACN,KAAK,iBAAiB,EACtB,0BAA0B,EAC1B,MAAM,mCAAmC,CAAA;AAC1C,OAAO,KAAK,EAAE,0BAA0B,EAAE,MAAM,sCAAsC,CAAA;AACtF,OAAO,KAAK,EAAE,wBAAwB,EAAE,MAAM,qCAAqC,CAAA;AAEnF,OAAO,KAAK,EAEX,sBAAsB,EACtB,MAAM,sCAAsC,CAAA;AAC7C,OAAO,KAAK,EACX,kBAAkB,EAClB,SAAS,EACT,gBAAgB,EAChB,cAAc,EACd,kBAAkB,EAClB,MAAM,2BAA2B,CAAA;AAGlC,OAAO,KAAK,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,0BAA0B,CAAA;AAClF,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,8BAA8B,CAAA;AAC3D,OAAO,EAAE,KAAK,WAAW,EAAgB,MAAM,iCAAiC,CAAA;AAChF,OAAO,KAAK,EAAY,gBAAgB,EAAE,MAAM,2BAA2B,CAAA;AAI3E,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAEhE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAMhE,OAAO,EAAE,KAAK,MAAM,EAAiB,MAAM,uBAAuB,CAAA;AAElE,OAAO,KAAK,
|
|
1
|
+
{"version":3,"file":"lifecycle.d.ts","sourceRoot":"","sources":["../../../src/manager/agent/lifecycle.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qCAAqC,CAAA;AACxE,OAAO,EACN,KAAK,iBAAiB,EACtB,0BAA0B,EAC1B,MAAM,mCAAmC,CAAA;AAC1C,OAAO,KAAK,EAAE,0BAA0B,EAAE,MAAM,sCAAsC,CAAA;AACtF,OAAO,KAAK,EAAE,wBAAwB,EAAE,MAAM,qCAAqC,CAAA;AAEnF,OAAO,KAAK,EAEX,sBAAsB,EACtB,MAAM,sCAAsC,CAAA;AAC7C,OAAO,KAAK,EACX,kBAAkB,EAClB,SAAS,EACT,gBAAgB,EAChB,cAAc,EACd,kBAAkB,EAClB,MAAM,2BAA2B,CAAA;AAGlC,OAAO,KAAK,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,0BAA0B,CAAA;AAClF,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,8BAA8B,CAAA;AAC3D,OAAO,EAAE,KAAK,WAAW,EAAgB,MAAM,iCAAiC,CAAA;AAChF,OAAO,KAAK,EAAY,gBAAgB,EAAE,MAAM,2BAA2B,CAAA;AAI3E,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAEhE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAMhE,OAAO,EAAE,KAAK,MAAM,EAAiB,MAAM,uBAAuB,CAAA;AAElE,OAAO,EAAE,KAAK,sBAAsB,EAAuB,MAAM,wBAAwB,CAAA;AAGzF;;;;;;;;;;;;;;;;;;GAkBG;AACH,UAAU,oBAAoB;IAC7B,QAAQ,CAAC,YAAY,EAAE,YAAY,CAAA;IACnC,QAAQ,CAAC,iBAAiB,EAAE,wBAAwB,CAAA;IACpD,QAAQ,CAAC,mBAAmB,EAAE,0BAA0B,CAAA;IACxD,QAAQ,CAAC,QAAQ,EAAE,iBAAiB,CAAA;IAEpC;;;;;;;;;;OAUG;IACH,QAAQ,CAAC,GAAG,CAAC,EAAE,MAAM,CAAA;CACrB;AAED;;;;;;;GAOG;AACH,MAAM,MAAM,gBAAgB,GAAG,oBAAoB,GAAG,sBAAsB,CAAA;AAE5E,UAAU,gBAAgB;IACzB,YAAY,EAAE,YAAY,CAAA;IAC1B,cAAc,EAAE,SAAS,CAAA;IACzB,QAAQ,EAAE,QAAQ,CAAA;IAClB,eAAe,EAAE,SAAS,CAAA;IAC1B,aAAa,EAAE,SAAS,CAAA;IACxB,UAAU,EAAE,MAAM,CAAA;IAClB,YAAY,CAAC,EAAE,YAAY,CAAA;IAC3B;;;;;;OAMG;IACH,kBAAkB,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CACtC;AAqBD,qBAAa,YAAY;IACxB,OAAO,CAAC,QAAQ,CAAe;IAC/B,OAAO,CAAC,SAAS,CAAoC;IACrD,OAAO,CAAC,YAAY,CAA2C;IAC/D,OAAO,CAAC,mBAAmB,CAA4C;IACvE,OAAO,CAAC,SAAS,CAA+B;IAChD,OAAO,CAAC,GAAG,CAAQ;IACnB,OAAO,CAAC,MAAM,CAA8B;IAC5C,OAAO,CAAC,cAAc,CAAwD;IAC9E,0EAA0E;IAC1E,OAAO,CAAC,UAAU,CAA2C;IAC7D,OAAO,CAAC,IAAI,CAAkB;IAC9B,OAAO,CAAC,YAAY,CAAc;gBAGjC,QAAQ,EAAE,aAAa,EACvB,MAAM,EAAE,OAAO,CAAC,kBAAkB,CAAC,GAAG,SAAS,EAC/C,IAAI,EAAE,gBAAgB;IAWjB,WAAW,CAChB,OAAO,EAAE,kBAAkB,EAC3B,OAAO,EAAE,gBAAgB,EACzB,QAAQ,CAAC,EAAE,gBAAgB,GACzB,OAAO,CAAC,SAAS,CAAC;IAuTrB,MAAM,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,CAAC,EAAE,WAAW,GAAG,IAAI;IAWjD,SAAS,CAAC,WAAW,EAAE,KAAK,EAAE,KAAK,GAAE,WAAsB,GAAG,IAAI;IAS5D,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAclE,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,GAAG,IAAI;IAapD,aAAa,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,EAAE;IAOxC,iBAAiB,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAehD,WAAW,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,GAAG,SAAS;IAIlD,cAAc,CAAC,MAAM,EAAE,MAAM,GAAG,gBAAgB,GAAG,SAAS;IAI5D,YAAY,CAAC,WAAW,EAAE,KAAK,GAAG,SAAS,EAAE;IAI7C,UAAU,IAAI,SAAS,EAAE;IAIzB,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,cAAc,GAAG,SAAS;IAIpD,WAAW,IAAI,aAAa;IAI5B,EAAE,CAAC,QAAQ,EAAE,sBAAsB,GAAG,IAAI;IAI1C,GAAG,CAAC,QAAQ,EAAE,sBAAsB,GAAG,IAAI;IAK3C,OAAO,IAAI,IAAI;IAUf,OAAO,IAAI,IAAI;IAgBf;;;;;;;;;;;;;OAaG;YACW,cAAc;YA0Bd,sBAAsB;YA6NtB,QAAQ;IAqBtB;;;;;OAKG;IACH,OAAO,CAAC,iBAAiB;YAuBX,aAAa;IAoG3B;;;;;;;;;;;;;;;;OAgBG;IACH,OAAO,CAAC,aAAa;IAgBrB,OAAO,CAAC,aAAa;IAoBrB,OAAO,CAAC,UAAU;YAiDJ,cAAc;IAc5B;;;;;;;;;;;;;OAaG;YACW,qBAAqB;IAiBnC,OAAO,CAAC,YAAY;IAkBpB,OAAO,CAAC,WAAW;IAOnB,OAAO,CAAC,eAAe;IAQvB,OAAO,CAAC,gBAAgB;IAgBxB,OAAO,CAAC,0BAA0B;IAQlC,OAAO,CAAC,kBAAkB;IAQ1B,OAAO,CAAC,IAAI;IAaZ,OAAO,CAAC,YAAY;CAYpB;AA+CD,OAAO,EAAE,0BAA0B,EAAE,CAAA"}
|
|
@@ -13,6 +13,7 @@ import { generateTaskId } from '../../utils/id.js';
|
|
|
13
13
|
import { SCOPE_ATTRIBUTE } from '../../utils/log/types.js';
|
|
14
14
|
import { resolveLogger } from '../../utils/logger.js';
|
|
15
15
|
import { requireOpenProject } from '../project/lifecycle.js';
|
|
16
|
+
import { resolveTopicManager } from '../topic/dependency.js';
|
|
16
17
|
/**
|
|
17
18
|
* Combine a child's own environment with what its parent passed down.
|
|
18
19
|
*
|
|
@@ -41,11 +42,15 @@ export class AgentManager {
|
|
|
41
42
|
/** One provisioning at a time per parent — see {@link provisionSpawn}. */
|
|
42
43
|
spawnLocks = new Map();
|
|
43
44
|
deps;
|
|
45
|
+
topicManager;
|
|
44
46
|
constructor(registry, config, deps) {
|
|
45
47
|
this.registry = registry;
|
|
46
48
|
this.config = { ...AGENT_MANAGER_DEFAULTS, ...config };
|
|
47
|
-
this.log = resolveLogger(deps.log).child({
|
|
49
|
+
this.log = resolveLogger(deps.log).child({
|
|
50
|
+
[SCOPE_ATTRIBUTE]: 'manager/agent/lifecycle',
|
|
51
|
+
});
|
|
48
52
|
this.deps = deps;
|
|
53
|
+
this.topicManager = resolveTopicManager(deps);
|
|
49
54
|
}
|
|
50
55
|
async sendMessage(options, context, listener) {
|
|
51
56
|
if (context.depth >= this.config.maxDepth) {
|
|
@@ -263,6 +268,8 @@ export class AgentManager {
|
|
|
263
268
|
model: options.configOverrides?.model ?? 'default',
|
|
264
269
|
tokenBudget: allocatedTokens,
|
|
265
270
|
timeoutMs: options.budgetAllocation?.timeoutMs ?? this.config.childTimeoutMs,
|
|
271
|
+
streamIdleTimeoutMs: options.configOverrides?.streamIdleTimeoutMs,
|
|
272
|
+
maxRequestRichContentBytes: options.configOverrides?.maxRequestRichContentBytes,
|
|
266
273
|
temperature: options.configOverrides?.temperature,
|
|
267
274
|
parentSpan: options.configOverrides?.parentSpan,
|
|
268
275
|
maxIterations: options.configOverrides?.maxIterations,
|
|
@@ -283,6 +290,12 @@ export class AgentManager {
|
|
|
283
290
|
resumeHandler: options.configOverrides?.resumeHandler ?? context.resumeHandler,
|
|
284
291
|
};
|
|
285
292
|
}
|
|
293
|
+
// Lineage is assigned by the spawning manager, not proposed by the
|
|
294
|
+
// child definition. A fixed configBuilder can ignore its inputs and
|
|
295
|
+
// configOverrides is caller-authored; neither may turn a child back
|
|
296
|
+
// into depth zero or attach it to a different parent run.
|
|
297
|
+
childConfig.parentRunId = context.parentRunId;
|
|
298
|
+
childConfig.depth = context.depth + 1;
|
|
286
299
|
// Stamped AFTER the builder, for the fourth time and the same reason:
|
|
287
300
|
// a `configBuilder` is written by whoever registered the agent and
|
|
288
301
|
// cannot forward a field it was never told about. A scope applied
|
|
@@ -505,7 +518,7 @@ export class AgentManager {
|
|
|
505
518
|
// SubSession + Session + WorkspaceRef triple (Convention #0: no
|
|
506
519
|
// partial/legacy path).
|
|
507
520
|
const store = this.deps.sessionStore;
|
|
508
|
-
//
|
|
521
|
+
// Topic archive gate — runs FIRST so an archived Topic fails fastest
|
|
509
522
|
// with the correct error (not DelegationCapacityExceeded or a project
|
|
510
523
|
// lookup error). Phase 2.6 closes the gap the Phase 2.5 commit
|
|
511
524
|
// flagged: without it, `TopicManager.archive` could be undermined by
|
|
@@ -515,7 +528,7 @@ export class AgentManager {
|
|
|
515
528
|
// callers of `SessionStore.createSession` bypass it — the store layer
|
|
516
529
|
// is intentionally unaware of topic status to preserve its
|
|
517
530
|
// single-responsibility boundary.
|
|
518
|
-
await this.
|
|
531
|
+
await this.topicManager.requireOpen(context.topicId, context.tenantId);
|
|
519
532
|
// Parent session cross-check: validate that `options.parentSessionId`
|
|
520
533
|
// exists for this tenant AND lives under the same topic as the
|
|
521
534
|
// context. A mismatched `context.topicId` would otherwise attach the
|