@hassangameryt/freeflow 0.2.0 → 0.4.0
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/.agents/plugins/marketplace.json +1 -1
- package/.claude-plugin/marketplace.json +4 -4
- package/{plugins/freeflow/.claude-plugin → .claude-plugin}/plugin.json +5 -2
- package/{plugins/freeflow/.codex-plugin → .codex-plugin}/plugin.json +8 -9
- package/README.md +194 -127
- package/assets/feedback-loop.png +0 -0
- package/assets/freeflow-plugin-icon.png +0 -0
- package/assets/freeflow_workflow_map.png +0 -0
- package/assets/workflow.png +0 -0
- package/command-surface.json +129 -0
- package/hooks/freeflow-runtime-context.mjs +419 -0
- package/hooks/tests/check-runtime-context-hook.sh +223 -0
- package/package.json +41 -22
- package/pi-extension/dist/index.js +269 -0
- package/pi-extension/dist/integrations/mcp-config.js +29 -0
- package/pi-extension/dist/output-router/host-producer-identification.js +232 -0
- package/pi-extension/dist/output-router/native-safety-net.js +216 -0
- package/pi-extension/dist/output-router/observed-tool-routing.js +122 -0
- package/pi-extension/dist/output-router/renderers.js +1023 -0
- package/pi-extension/dist/output-router/router-tools.js +554 -0
- package/pi-extension/dist/output-router/schemas.js +383 -0
- package/pi-extension/dist/output-router/status.js +744 -0
- package/pi-extension/dist/output-router/utils.js +502 -0
- package/pi-extension/dist/runtime/runtime-context.js +798 -0
- package/pi-extension/dist/settings/settings-tui.js +314 -0
- package/pi-extension/dist/settings/settings-ui.js +1487 -0
- package/pi-extension/freeflow/index.js +1 -0
- package/pi-extension/src/index.ts +295 -0
- package/pi-extension/src/integrations/mcp-config.ts +32 -0
- package/pi-extension/src/output-router/host-producer-identification.ts +256 -0
- package/pi-extension/src/output-router/native-safety-net.ts +248 -0
- package/pi-extension/src/output-router/observed-tool-routing.ts +134 -0
- package/pi-extension/src/output-router/renderers.ts +1108 -0
- package/pi-extension/src/output-router/router-tools.ts +579 -0
- package/pi-extension/src/output-router/schemas.ts +387 -0
- package/pi-extension/src/output-router/status.ts +791 -0
- package/pi-extension/src/output-router/utils.ts +554 -0
- package/pi-extension/src/runtime/runtime-context.ts +895 -0
- package/pi-extension/src/settings/settings-tui.ts +408 -0
- package/pi-extension/src/settings/settings-ui.ts +1730 -0
- package/pi-extension/tests/integration/pi-extension.test.js +4018 -0
- package/pi-extension/tests/output-router/observed-producers.test.js +132 -0
- package/pi-extension/tests/output-router/search.test.js +722 -0
- package/pi-extension/tests/settings/settings-tui.test.js +243 -0
- package/pi-extension/tsconfig.json +16 -0
- package/{plugins/freeflow/router/dist → router/dist/benchmarks}/benchmark-harness.d.ts +16 -13
- package/router/dist/benchmarks/benchmark-harness.js +92 -0
- package/router/dist/benchmarks/benchmarks.d.ts +109 -0
- package/router/dist/benchmarks/benchmarks.js +979 -0
- package/router/dist/benchmarks/codex-qa-benchmarks.d.ts +95 -0
- package/router/dist/benchmarks/codex-qa-benchmarks.js +500 -0
- package/router/dist/benchmarks/command-benchmarks.d.ts +143 -0
- package/router/dist/benchmarks/command-benchmarks.js +698 -0
- package/router/dist/benchmarks/context-mode-normalized-benchmarks.d.ts +82 -0
- package/router/dist/benchmarks/context-mode-normalized-benchmarks.js +773 -0
- package/router/dist/benchmarks/context-mode-real-deep-benchmark.d.ts +132 -0
- package/router/dist/benchmarks/context-mode-real-deep-benchmark.js +1380 -0
- package/router/dist/benchmarks/context-saving-baseline.d.ts +60 -0
- package/router/dist/benchmarks/context-saving-baseline.js +470 -0
- package/router/dist/benchmarks/index-benchmarks.d.ts +103 -0
- package/router/dist/benchmarks/index-benchmarks.js +862 -0
- package/router/dist/benchmarks/storage-policy-benchmarks.d.ts +101 -0
- package/router/dist/benchmarks/storage-policy-benchmarks.js +618 -0
- package/router/dist/benchmarks/vault-index-benchmarks.d.ts +56 -0
- package/router/dist/benchmarks/vault-index-benchmarks.js +322 -0
- package/router/dist/config/config.d.ts +103 -0
- package/router/dist/config/config.js +632 -0
- package/{plugins/freeflow/router/dist → router/dist/config}/router-contract.d.ts +2 -2
- package/router/dist/config/router-contract.js +54 -0
- package/{plugins/freeflow/router/dist → router/dist/config}/schema.d.ts +19 -10
- package/router/dist/config/schema.js +648 -0
- package/router/dist/config/types.d.ts +573 -0
- package/router/dist/config/types.js +41 -0
- package/router/dist/evidence/bounded-evidence.d.ts +40 -0
- package/router/dist/evidence/bounded-evidence.js +323 -0
- package/router/dist/evidence/evidence-range-selector.d.ts +24 -0
- package/router/dist/evidence/evidence-range-selector.js +232 -0
- package/router/dist/evidence/evidence-search.d.ts +27 -0
- package/router/dist/evidence/evidence-search.js +490 -0
- package/{plugins/freeflow/router/dist → router/dist/evidence}/evidence.d.ts +18 -18
- package/router/dist/evidence/evidence.js +200 -0
- package/router/dist/evidence/failure-contracts.d.ts +37 -0
- package/router/dist/evidence/failure-contracts.js +166 -0
- package/router/dist/evidence/line-ranges.d.ts +24 -0
- package/router/dist/evidence/line-ranges.js +22 -0
- package/router/dist/experiments/local-index.d.ts +64 -0
- package/router/dist/experiments/local-index.js +581 -0
- package/router/dist/index.d.ts +30 -0
- package/router/dist/index.js +32 -0
- package/router/dist/local/local-traversal.d.ts +18 -0
- package/router/dist/local/local-traversal.js +334 -0
- package/router/dist/processing/engine.d.ts +132 -0
- package/router/dist/processing/engine.js +533 -0
- package/router/dist/processing/reducers.d.ts +280 -0
- package/router/dist/processing/reducers.js +1673 -0
- package/router/dist/processing/renderers.d.ts +31 -0
- package/router/dist/processing/renderers.js +195 -0
- package/router/dist/processing/scripts.d.ts +91 -0
- package/router/dist/processing/scripts.js +386 -0
- package/{plugins/freeflow/router/dist → router/dist/repo}/repo-traversal.d.ts +9 -9
- package/router/dist/repo/repo-traversal.js +245 -0
- package/router/dist/routing/observed-reducers.d.ts +41 -0
- package/router/dist/routing/observed-reducers.js +661 -0
- package/router/dist/routing/observed-routing.d.ts +26 -0
- package/router/dist/routing/observed-routing.js +378 -0
- package/router/dist/routing/parsers.d.ts +16 -0
- package/router/dist/routing/parsers.js +521 -0
- package/router/dist/routing/run-filters.d.ts +60 -0
- package/router/dist/routing/run-filters.js +238 -0
- package/router/dist/routing/run-reducers.d.ts +49 -0
- package/router/dist/routing/run-reducers.js +122 -0
- package/router/dist/sandbox/adapter-roots.d.ts +46 -0
- package/router/dist/sandbox/adapter-roots.js +66 -0
- package/router/dist/sandbox/eryx-python-adapter.d.ts +13 -0
- package/router/dist/sandbox/eryx-python-adapter.js +868 -0
- package/router/dist/sandbox/jq-wasm-adapter.d.ts +9 -0
- package/router/dist/sandbox/jq-wasm-adapter.js +429 -0
- package/router/dist/sandbox/quickjs-wasi-adapter.d.ts +13 -0
- package/router/dist/sandbox/quickjs-wasi-adapter.js +384 -0
- package/router/dist/sandbox/script-sandbox.d.ts +134 -0
- package/router/dist/sandbox/script-sandbox.js +337 -0
- package/router/dist/setup/script-transform-adapters.d.ts +32 -0
- package/router/dist/setup/script-transform-adapters.js +278 -0
- package/router/dist/tools/batch.d.ts +33 -0
- package/router/dist/tools/batch.js +586 -0
- package/router/dist/tools/run.d.ts +62 -0
- package/router/dist/tools/run.js +1304 -0
- package/router/dist/tools/search.d.ts +66 -0
- package/router/dist/tools/search.js +1364 -0
- package/router/dist/transform/engine.d.ts +168 -0
- package/router/dist/transform/engine.js +2072 -0
- package/router/dist/vault/vault-index.d.ts +125 -0
- package/router/dist/vault/vault-index.js +416 -0
- package/router/dist/vault/vault.d.ts +138 -0
- package/router/dist/vault/vault.js +630 -0
- package/runtime/interaction-contract.md +16 -0
- package/skills/bypass/SKILL.md +82 -0
- package/skills/commit-work/SKILL.md +106 -0
- package/skills/commit-work/references/staging-decisions.md +73 -0
- package/skills/decision-gate/SKILL.md +48 -0
- package/skills/design-for-depth/SKILL.md +106 -0
- package/skills/design-for-depth/references/design-pressure-signals.md +314 -0
- package/skills/design-for-depth/references/interface-design-loop.md +156 -0
- package/skills/design-for-depth/references/software-design-philosophy.md +248 -0
- package/skills/diagnose-failure/SKILL.md +105 -0
- package/skills/diagnose-failure/references/diagnostic-loop-catalog.md +36 -0
- package/skills/diagnose-failure/references/flaky-and-performance.md +43 -0
- package/skills/discuss/SKILL.md +121 -0
- package/skills/discuss/references/checkpoints.md +66 -0
- package/skills/evaluate-skill/SKILL.md +105 -0
- package/skills/evaluate-skill/references/definition-schema.md +345 -0
- package/skills/evaluate-skill/references/evaluation-design.md +86 -0
- package/skills/evaluate-skill/references/execution-and-evidence.md +85 -0
- package/skills/evaluate-skill/references/review-and-revision.md +51 -0
- package/skills/evaluate-skill/scripts/lib/definitions.mjs +296 -0
- package/skills/evaluate-skill/scripts/lib/evidence.mjs +46 -0
- package/skills/evaluate-skill/scripts/lib/grade-comparison.mjs +60 -0
- package/skills/evaluate-skill/scripts/lib/grade-interaction.mjs +148 -0
- package/skills/evaluate-skill/scripts/lib/grade-support.mjs +126 -0
- package/skills/evaluate-skill/scripts/lib/grade-workspace.mjs +192 -0
- package/skills/evaluate-skill/scripts/lib/grade.mjs +56 -0
- package/skills/evaluate-skill/scripts/lib/pi.mjs +489 -0
- package/skills/evaluate-skill/scripts/lib/process.mjs +79 -0
- package/skills/evaluate-skill/scripts/lib/rpc.mjs +444 -0
- package/skills/evaluate-skill/scripts/lib/runner.mjs +346 -0
- package/skills/evaluate-skill/scripts/lib/sandbox.mjs +409 -0
- package/skills/evaluate-skill/scripts/lib/view.mjs +337 -0
- package/skills/evaluate-skill/scripts/pi-guard.mjs +164 -0
- package/skills/evaluate-skill/scripts/skill-eval.mjs +101 -0
- package/skills/execute-work/SKILL.md +99 -0
- package/skills/execute-work/references/code-practices.md +78 -0
- package/skills/execute-work/references/execution-loop.md +77 -0
- package/skills/finish-branch/SKILL.md +117 -0
- package/skills/finish-branch/references/integration-options.md +80 -0
- package/skills/handoff/SKILL.md +92 -0
- package/skills/handoff/references/templates.md +141 -0
- package/skills/launch-work/SKILL.md +116 -0
- package/skills/launch-work/references/launch-readiness.md +85 -0
- package/skills/migration-work/SKILL.md +115 -0
- package/skills/migration-work/references/migration-lifecycle.md +82 -0
- package/skills/mode-contract/SKILL.md +82 -0
- package/skills/output-router/SKILL.md +206 -0
- package/skills/output-router/references/safety-policy.md +40 -0
- package/skills/release-work/SKILL.md +118 -0
- package/skills/release-work/references/release-evidence.md +78 -0
- package/skills/review-artifact/SKILL.md +126 -0
- package/skills/review-artifact/references/reviewer-prompt.md +148 -0
- package/skills/review-work/SKILL.md +126 -0
- package/skills/review-work/references/reviewer-prompt.md +134 -0
- package/skills/review-work/references/security-risk-lens.md +67 -0
- package/skills/setup-freeflow/SKILL.md +98 -0
- package/skills/setup-freeflow/references/activation-contract.md +120 -0
- package/skills/setup-freeflow/references/host-setup.md +72 -0
- package/skills/setup-freeflow/references/output-router-setup.md +110 -0
- package/skills/simplify-code/SKILL.md +80 -0
- package/skills/simplify-code/references/simplification-patterns.md +70 -0
- package/skills/tdd/SKILL.md +96 -0
- package/skills/tdd/references/test-design.md +96 -0
- package/skills/track-work/SKILL.md +84 -0
- package/skills/track-work/references/working-record-schema.md +378 -0
- package/skills/verify-work/SKILL.md +85 -0
- package/skills/verify-work/references/browser-runtime-evidence.md +55 -0
- package/skills/verify-work/references/integration-evidence.md +51 -0
- package/skills/verify-work/references/performance-evidence.md +57 -0
- package/skills/workflow/SKILL.md +91 -0
- package/skills/workflow/references/domain-skill-composition.md +57 -0
- package/skills/workflow/references/workflow-loop.md +92 -0
- package/skills/write-plan/SKILL.md +123 -0
- package/skills/write-plan/references/plan-shapes.md +156 -0
- package/skills/write-skill/SKILL.md +119 -0
- package/skills/write-skill/references/activation-boundaries.md +62 -0
- package/skills/write-skill/references/agent-first-instructions.md +87 -0
- package/skills/write-skill/references/progressive-disclosure.md +63 -0
- package/skills/write-skill/scripts/lib/skill-author.mjs +653 -0
- package/skills/write-skill/scripts/skill-author.mjs +104 -0
- package/skills/write-spec/SKILL.md +104 -0
- package/skills/write-spec/references/artifact-standards.md +86 -0
- package/skills/write-spec/references/decision-records.md +77 -0
- package/skills/write-spec/references/spec-shapes.md +128 -0
- package/plugins/freeflow/command-surface.json +0 -77
- package/plugins/freeflow/docs/README.md +0 -40
- package/plugins/freeflow/docs/adr/0001-three-modes.md +0 -27
- package/plugins/freeflow/docs/adr/0002-evals-before-hooks.md +0 -17
- package/plugins/freeflow/docs/adr/0003-release-boundary.md +0 -21
- package/plugins/freeflow/docs/adr/0004-research-replaces-shallow-discovery-skills.md +0 -26
- package/plugins/freeflow/docs/adr/README.md +0 -8
- package/plugins/freeflow/docs/architecture.md +0 -86
- package/plugins/freeflow/docs/output-router.md +0 -212
- package/plugins/freeflow/docs/release-evidence.md +0 -69
- package/plugins/freeflow/docs/skills.md +0 -43
- package/plugins/freeflow/docs/workflow.md +0 -57
- package/plugins/freeflow/hooks/freeflow-runtime-context.mjs +0 -257
- package/plugins/freeflow/pi-extension/index.js +0 -1176
- package/plugins/freeflow/router/dist/benchmark-harness.js +0 -92
- package/plugins/freeflow/router/dist/benchmarks.d.ts +0 -102
- package/plugins/freeflow/router/dist/benchmarks.js +0 -880
- package/plugins/freeflow/router/dist/bounded-evidence.d.ts +0 -40
- package/plugins/freeflow/router/dist/bounded-evidence.js +0 -310
- package/plugins/freeflow/router/dist/codex-qa-benchmarks.d.ts +0 -91
- package/plugins/freeflow/router/dist/codex-qa-benchmarks.js +0 -461
- package/plugins/freeflow/router/dist/command-benchmarks.d.ts +0 -128
- package/plugins/freeflow/router/dist/command-benchmarks.js +0 -652
- package/plugins/freeflow/router/dist/config.d.ts +0 -22
- package/plugins/freeflow/router/dist/config.js +0 -114
- package/plugins/freeflow/router/dist/evidence-range-selector.d.ts +0 -22
- package/plugins/freeflow/router/dist/evidence-range-selector.js +0 -180
- package/plugins/freeflow/router/dist/evidence-search.d.ts +0 -25
- package/plugins/freeflow/router/dist/evidence-search.js +0 -450
- package/plugins/freeflow/router/dist/evidence.js +0 -196
- package/plugins/freeflow/router/dist/experimental-local-index.d.ts +0 -1
- package/plugins/freeflow/router/dist/experimental-local-index.js +0 -1
- package/plugins/freeflow/router/dist/experiments/local-index.d.ts +0 -58
- package/plugins/freeflow/router/dist/experiments/local-index.js +0 -562
- package/plugins/freeflow/router/dist/index-benchmarks.d.ts +0 -84
- package/plugins/freeflow/router/dist/index-benchmarks.js +0 -458
- package/plugins/freeflow/router/dist/index.d.ts +0 -12
- package/plugins/freeflow/router/dist/index.js +0 -12
- package/plugins/freeflow/router/dist/line-ranges.d.ts +0 -22
- package/plugins/freeflow/router/dist/line-ranges.js +0 -22
- package/plugins/freeflow/router/dist/parsers.d.ts +0 -16
- package/plugins/freeflow/router/dist/parsers.js +0 -473
- package/plugins/freeflow/router/dist/repo-traversal.js +0 -242
- package/plugins/freeflow/router/dist/retrieve.d.ts +0 -33
- package/plugins/freeflow/router/dist/retrieve.js +0 -855
- package/plugins/freeflow/router/dist/router-contract.js +0 -54
- package/plugins/freeflow/router/dist/run.d.ts +0 -26
- package/plugins/freeflow/router/dist/run.js +0 -356
- package/plugins/freeflow/router/dist/schema.js +0 -354
- package/plugins/freeflow/router/dist/types.d.ts +0 -234
- package/plugins/freeflow/router/dist/types.js +0 -9
- package/plugins/freeflow/router/dist/vault.d.ts +0 -80
- package/plugins/freeflow/router/dist/vault.js +0 -484
- package/plugins/freeflow/skills/bypass/SKILL.md +0 -68
- package/plugins/freeflow/skills/commit-work/SKILL.md +0 -98
- package/plugins/freeflow/skills/commit-work/references/staging-decisions.md +0 -51
- package/plugins/freeflow/skills/diagnose-failure/SKILL.md +0 -95
- package/plugins/freeflow/skills/diagnose-failure/references/feedback-loop-catalog.md +0 -25
- package/plugins/freeflow/skills/diagnose-failure/references/flaky-and-performance.md +0 -31
- package/plugins/freeflow/skills/evaluate-skill/SKILL.md +0 -96
- package/plugins/freeflow/skills/evaluate-skill/references/eval-patterns.md +0 -41
- package/plugins/freeflow/skills/evaluate-skill/references/grading-priority.md +0 -33
- package/plugins/freeflow/skills/execute-plan/SKILL.md +0 -132
- package/plugins/freeflow/skills/execute-plan/references/execution-map.md +0 -122
- package/plugins/freeflow/skills/handoff/SKILL.md +0 -66
- package/plugins/freeflow/skills/handoff/references/templates.md +0 -64
- package/plugins/freeflow/skills/interview-gate/SKILL.md +0 -90
- package/plugins/freeflow/skills/mode-contract/SKILL.md +0 -79
- package/plugins/freeflow/skills/output-router/SKILL.md +0 -69
- package/plugins/freeflow/skills/output-router/references/safety-policy.md +0 -33
- package/plugins/freeflow/skills/research/ARTIFACT-DESTINATIONS.md +0 -58
- package/plugins/freeflow/skills/research/CHECKPOINTS.md +0 -74
- package/plugins/freeflow/skills/research/LANGUAGE.md +0 -61
- package/plugins/freeflow/skills/research/SKILL.md +0 -128
- package/plugins/freeflow/skills/review-artifact/SKILL.md +0 -115
- package/plugins/freeflow/skills/review-artifact/references/reviewer-prompt.md +0 -56
- package/plugins/freeflow/skills/review-work/SKILL.md +0 -96
- package/plugins/freeflow/skills/review-work/references/reviewer-prompt.md +0 -97
- package/plugins/freeflow/skills/setup-freeflow/SKILL.md +0 -94
- package/plugins/freeflow/skills/setup-freeflow/references/activation-contract.md +0 -99
- package/plugins/freeflow/skills/setup-freeflow/references/host-setup.md +0 -35
- package/plugins/freeflow/skills/setup-freeflow/references/output-router-setup.md +0 -52
- package/plugins/freeflow/skills/verify-work/SKILL.md +0 -62
- package/plugins/freeflow/skills/workflow/SKILL.md +0 -95
- package/plugins/freeflow/skills/workflow/references/workflow-map.md +0 -102
- package/plugins/freeflow/skills/write-plan/SKILL.md +0 -88
- package/plugins/freeflow/skills/write-plan/references/plan-shapes.md +0 -70
- package/plugins/freeflow/skills/write-skill/SKILL.md +0 -88
- package/plugins/freeflow/skills/write-spec/SKILL.md +0 -90
- package/plugins/freeflow/skills/write-spec/references/artifact-standards.md +0 -67
- package/plugins/freeflow/skills/write-spec/references/spec-shapes.md +0 -69
- /package/{plugins/freeflow/assets → assets}/freeflow-plugin-icon-transparent.png +0 -0
- /package/{plugins/freeflow/hooks → hooks}/hooks.json +0 -0
|
@@ -0,0 +1,346 @@
|
|
|
1
|
+
import { mkdir, realpath } from "node:fs/promises";
|
|
2
|
+
import path from "node:path";
|
|
3
|
+
|
|
4
|
+
import { loadDefinition, selectDefinition } from "./definitions.mjs";
|
|
5
|
+
import { createInvocationId, fileIdentity, writeJson, writeText } from "./evidence.mjs";
|
|
6
|
+
import { gradeDeterministic } from "./grade.mjs";
|
|
7
|
+
import { runBody, runDescription, VariantSetupError } from "./pi.mjs";
|
|
8
|
+
import { prepareFixture } from "./sandbox.mjs";
|
|
9
|
+
|
|
10
|
+
const VARIANTS = ["baseline", "candidate"];
|
|
11
|
+
const DESCRIPTION_TOOLS = new Set(["read"]);
|
|
12
|
+
const BODY_TOOLS = new Set(["read", "write", "edit"]);
|
|
13
|
+
|
|
14
|
+
/**
|
|
15
|
+
* @param {string} definitionFile
|
|
16
|
+
* @param {any} selectors
|
|
17
|
+
* @param {{root?: string, signal?: AbortSignal}} options
|
|
18
|
+
*/
|
|
19
|
+
export async function runEvaluation(definitionFile, selectors = {}, { root = process.cwd(), signal } = {}) {
|
|
20
|
+
const definitionPath = await realpath(path.resolve(root, definitionFile));
|
|
21
|
+
const definition = await loadDefinition(definitionPath, { root });
|
|
22
|
+
const selection = selectDefinition(definition, selectors);
|
|
23
|
+
assertSupportedSelection(selection);
|
|
24
|
+
const invocationId = createInvocationId();
|
|
25
|
+
const runsDirectory = path.join(root, ".skill-eval", "runs");
|
|
26
|
+
const resultDirectory = path.join(runsDirectory, invocationId);
|
|
27
|
+
await mkdir(runsDirectory, { recursive: true });
|
|
28
|
+
await mkdir(resultDirectory);
|
|
29
|
+
await writeJson(path.join(resultDirectory, "invocation.json"), {
|
|
30
|
+
schema_version: 1,
|
|
31
|
+
id: invocationId,
|
|
32
|
+
definition: definitionPath,
|
|
33
|
+
selection: {
|
|
34
|
+
group: selectors.group ?? null,
|
|
35
|
+
variant: selectors.variant ?? null,
|
|
36
|
+
groups: selection.groups.map((entry) => ({
|
|
37
|
+
id: entry.group.id,
|
|
38
|
+
position: entry.position,
|
|
39
|
+
variants: entry.variants,
|
|
40
|
+
})),
|
|
41
|
+
},
|
|
42
|
+
startedAt: new Date().toISOString(),
|
|
43
|
+
});
|
|
44
|
+
|
|
45
|
+
const groups = [];
|
|
46
|
+
for (const selected of selection.groups) {
|
|
47
|
+
groups.push(await runGroup({ selected, root, resultDirectory, signal }));
|
|
48
|
+
}
|
|
49
|
+
const state = batchState(groups);
|
|
50
|
+
const summary = {
|
|
51
|
+
schema_version: 1,
|
|
52
|
+
id: invocationId,
|
|
53
|
+
state,
|
|
54
|
+
definitionKind: selection.definitionKind,
|
|
55
|
+
groups,
|
|
56
|
+
completedAt: new Date().toISOString(),
|
|
57
|
+
};
|
|
58
|
+
await writeJson(path.join(resultDirectory, "summary.json"), summary);
|
|
59
|
+
return { id: invocationId, path: resultDirectory, state, summary };
|
|
60
|
+
}
|
|
61
|
+
|
|
62
|
+
function assertSupportedSelection(selection) {
|
|
63
|
+
for (const selected of selection.groups) {
|
|
64
|
+
const { group } = selected;
|
|
65
|
+
if (group.type !== "description" && group.type !== "body") {
|
|
66
|
+
throw new Error("skill-eval run currently supports description and body groups only");
|
|
67
|
+
}
|
|
68
|
+
if (group.expectations.some((expectation) => expectation.kind === "command")) {
|
|
69
|
+
throw new Error("command expectations are not supported");
|
|
70
|
+
}
|
|
71
|
+
const supportedTools = group.type === "body" ? BODY_TOOLS : DESCRIPTION_TOOLS;
|
|
72
|
+
if (group.tools.some((tool) => !supportedTools.has(tool))) {
|
|
73
|
+
throw new Error(`${group.type} execution currently supports only ${[...supportedTools].join(", ")} tools`);
|
|
74
|
+
}
|
|
75
|
+
}
|
|
76
|
+
}
|
|
77
|
+
|
|
78
|
+
async function runGroup({ selected, root, resultDirectory, signal }) {
|
|
79
|
+
const { group } = selected;
|
|
80
|
+
const groupDirectory = path.join(resultDirectory, "groups", group.id);
|
|
81
|
+
await mkdir(groupDirectory, { recursive: true });
|
|
82
|
+
await writeJson(path.join(groupDirectory, "definition.json"), group);
|
|
83
|
+
|
|
84
|
+
let fixture = null;
|
|
85
|
+
let fixtureError = null;
|
|
86
|
+
if (!signal?.aborted) {
|
|
87
|
+
try {
|
|
88
|
+
fixture = await prepareFixture({ declaredPath: group.fixture, root, groupDirectory });
|
|
89
|
+
} catch (error) {
|
|
90
|
+
fixtureError = error instanceof Error ? error.message : String(error);
|
|
91
|
+
}
|
|
92
|
+
}
|
|
93
|
+
|
|
94
|
+
const runs = {};
|
|
95
|
+
const runFiles = {};
|
|
96
|
+
for (const variant of VARIANTS) {
|
|
97
|
+
const variantDirectory = path.join(groupDirectory, variant);
|
|
98
|
+
await mkdir(variantDirectory, { recursive: true });
|
|
99
|
+
if (selected.variants[variant] === "not-selected") {
|
|
100
|
+
runs[variant] = notSelectedRun(group, variant);
|
|
101
|
+
} else if (signal?.aborted) {
|
|
102
|
+
runs[variant] = cancelledRun(group, variant);
|
|
103
|
+
} else if (fixtureError !== null) {
|
|
104
|
+
runs[variant] = invalidRun(group, variant, fixtureError);
|
|
105
|
+
} else {
|
|
106
|
+
runs[variant] = await executeVariant({ group, variant, root, variantDirectory, fixture, signal });
|
|
107
|
+
}
|
|
108
|
+
const persisted = await persistRunOrFailure(variantDirectory, runs[variant]);
|
|
109
|
+
runs[variant] = persisted.run;
|
|
110
|
+
runFiles[variant] = persisted.file;
|
|
111
|
+
}
|
|
112
|
+
|
|
113
|
+
const evidence = {
|
|
114
|
+
baseline: await fileIdentity(runFiles.baseline),
|
|
115
|
+
candidate: await fileIdentity(runFiles.candidate),
|
|
116
|
+
};
|
|
117
|
+
const grade = await createGrade(group, runs, evidence);
|
|
118
|
+
const persistedGrade = await persistGrade(groupDirectory, grade);
|
|
119
|
+
const selectedRuns = VARIANTS.filter((variant) => selected.variants[variant] === "selected").map(
|
|
120
|
+
(variant) => runs[variant],
|
|
121
|
+
);
|
|
122
|
+
const selectedState = selectedGroupState(selectedRuns, persistedGrade.grade);
|
|
123
|
+
const groupState = selectedState === "cancelled" ? selectedState : fixtureError === null ? selectedState : "invalid";
|
|
124
|
+
const groupResult = {
|
|
125
|
+
id: group.id,
|
|
126
|
+
position: selected.position,
|
|
127
|
+
state: groupState,
|
|
128
|
+
variants: Object.fromEntries(VARIANTS.map((variant) => [variant, runs[variant].state])),
|
|
129
|
+
grade: persistedGrade.grade.state,
|
|
130
|
+
artifacts: { grade: persistedGrade.artifact, group: "group.json" },
|
|
131
|
+
errors: persistedGrade.errors,
|
|
132
|
+
};
|
|
133
|
+
return persistGroupResult(groupDirectory, groupResult);
|
|
134
|
+
}
|
|
135
|
+
|
|
136
|
+
function batchState(groups) {
|
|
137
|
+
if (groups.some((group) => group.state === "cancelled")) return "cancelled";
|
|
138
|
+
if (groups.every((group) => group.state === "complete")) return "complete";
|
|
139
|
+
return "partially-complete";
|
|
140
|
+
}
|
|
141
|
+
|
|
142
|
+
function selectedGroupState(runs, grade) {
|
|
143
|
+
if (runs.some((run) => run.state === "cancelled")) return "cancelled";
|
|
144
|
+
if (runs.every((run) => run.state === "complete") && grade.state !== "grade-error") return "complete";
|
|
145
|
+
return "partially-complete";
|
|
146
|
+
}
|
|
147
|
+
|
|
148
|
+
async function createGrade(group, runs, evidence) {
|
|
149
|
+
try {
|
|
150
|
+
return await gradeDeterministic(group, runs, evidence);
|
|
151
|
+
} catch (error) {
|
|
152
|
+
return gradeErrorReport(evidence, `deterministic grading failed: ${errorMessage(error)}`);
|
|
153
|
+
}
|
|
154
|
+
}
|
|
155
|
+
|
|
156
|
+
async function persistGrade(groupDirectory, grade) {
|
|
157
|
+
const artifact = "deterministic-grade.json";
|
|
158
|
+
try {
|
|
159
|
+
await writeJson(path.join(groupDirectory, artifact), grade);
|
|
160
|
+
return { grade, artifact, errors: [] };
|
|
161
|
+
} catch (error) {
|
|
162
|
+
const reason = `failed to persist ${artifact}: ${errorMessage(error)}`;
|
|
163
|
+
const failed = gradeErrorReport(grade.evidence, reason, grade);
|
|
164
|
+
const fallback = "deterministic-grade-error.json";
|
|
165
|
+
await writeJson(path.join(groupDirectory, fallback), failed);
|
|
166
|
+
return {
|
|
167
|
+
grade: failed,
|
|
168
|
+
artifact: fallback,
|
|
169
|
+
errors: [{ kind: "persistence", artifact, message: reason }],
|
|
170
|
+
};
|
|
171
|
+
}
|
|
172
|
+
}
|
|
173
|
+
|
|
174
|
+
async function persistGroupResult(groupDirectory, groupResult) {
|
|
175
|
+
const artifact = "group.json";
|
|
176
|
+
try {
|
|
177
|
+
await writeJson(path.join(groupDirectory, artifact), groupResult);
|
|
178
|
+
return groupResult;
|
|
179
|
+
} catch (error) {
|
|
180
|
+
const message = `failed to persist ${artifact}: ${errorMessage(error)}`;
|
|
181
|
+
const fallback = "group-error.json";
|
|
182
|
+
const failed = {
|
|
183
|
+
...groupResult,
|
|
184
|
+
state: groupResult.state === "complete" ? "partially-complete" : groupResult.state,
|
|
185
|
+
artifacts: { ...groupResult.artifacts, group: fallback },
|
|
186
|
+
errors: [...groupResult.errors, { kind: "persistence", artifact, message }],
|
|
187
|
+
};
|
|
188
|
+
await writeJson(path.join(groupDirectory, fallback), failed);
|
|
189
|
+
return failed;
|
|
190
|
+
}
|
|
191
|
+
}
|
|
192
|
+
|
|
193
|
+
function gradeErrorReport(evidence, reason, grade = null) {
|
|
194
|
+
return {
|
|
195
|
+
schema_version: 1,
|
|
196
|
+
state: "grade-error",
|
|
197
|
+
evidence,
|
|
198
|
+
checks: grade?.checks ?? [],
|
|
199
|
+
comparisons: grade?.comparisons ?? [],
|
|
200
|
+
errors: [...(grade?.errors ?? []), { id: null, reason }],
|
|
201
|
+
};
|
|
202
|
+
}
|
|
203
|
+
|
|
204
|
+
function errorMessage(error) {
|
|
205
|
+
return error instanceof Error ? error.message : String(error);
|
|
206
|
+
}
|
|
207
|
+
|
|
208
|
+
async function executeVariant({ group, variant, root, variantDirectory, fixture, signal }) {
|
|
209
|
+
try {
|
|
210
|
+
const run = group.type === "body" ? runBody : runDescription;
|
|
211
|
+
return await run({ group, variant, root, variantDirectory, fixture, signal });
|
|
212
|
+
} catch (error) {
|
|
213
|
+
if (error instanceof VariantSetupError) return invalidRun(group, variant, error.message);
|
|
214
|
+
return infrastructureFailedRun(group, variant, error);
|
|
215
|
+
}
|
|
216
|
+
}
|
|
217
|
+
|
|
218
|
+
async function persistRunOrFailure(variantDirectory, run) {
|
|
219
|
+
try {
|
|
220
|
+
return { run, file: await persistRun(variantDirectory, run) };
|
|
221
|
+
} catch (error) {
|
|
222
|
+
const failed = persistenceFailedRun(run, error);
|
|
223
|
+
const runFile = path.join(variantDirectory, "run.json");
|
|
224
|
+
await writeJson(runFile, {
|
|
225
|
+
...failed,
|
|
226
|
+
transcript: failed.transcript ?? [],
|
|
227
|
+
artifacts: {
|
|
228
|
+
events: failed.process === undefined ? null : "events.jsonl",
|
|
229
|
+
final: null,
|
|
230
|
+
stderr: failed.process === undefined ? null : "stderr.log",
|
|
231
|
+
},
|
|
232
|
+
});
|
|
233
|
+
return { run: failed, file: runFile };
|
|
234
|
+
}
|
|
235
|
+
}
|
|
236
|
+
|
|
237
|
+
async function persistRun(variantDirectory, run) {
|
|
238
|
+
const transcript = run.transcript ?? [];
|
|
239
|
+
const response = run.response ?? "";
|
|
240
|
+
await writeJson(path.join(variantDirectory, "transcript.json"), transcript);
|
|
241
|
+
await writeText(path.join(variantDirectory, "final.md"), response === "" ? "" : `${response}\n`);
|
|
242
|
+
const runFile = path.join(variantDirectory, "run.json");
|
|
243
|
+
const hasProcessEvidence = run.process !== undefined;
|
|
244
|
+
await writeJson(runFile, {
|
|
245
|
+
...run,
|
|
246
|
+
...(Array.isArray(run.turns) ? { turns: externalizeTurnTranscripts(run.turns) } : {}),
|
|
247
|
+
transcript: "transcript.json",
|
|
248
|
+
artifacts: {
|
|
249
|
+
events: hasProcessEvidence ? "events.jsonl" : null,
|
|
250
|
+
final: "final.md",
|
|
251
|
+
stderr: hasProcessEvidence ? "stderr.log" : null,
|
|
252
|
+
},
|
|
253
|
+
});
|
|
254
|
+
return runFile;
|
|
255
|
+
}
|
|
256
|
+
|
|
257
|
+
function externalizeTurnTranscripts(turns) {
|
|
258
|
+
let offset = 0;
|
|
259
|
+
return turns.map((turn) => {
|
|
260
|
+
const length = Array.isArray(turn.transcript) ? turn.transcript.length : 0;
|
|
261
|
+
const transcript = { file: "transcript.json", start: offset, end: offset + length };
|
|
262
|
+
offset += length;
|
|
263
|
+
return { ...turn, transcript };
|
|
264
|
+
});
|
|
265
|
+
}
|
|
266
|
+
|
|
267
|
+
function notSelectedRun(group, variant) {
|
|
268
|
+
return {
|
|
269
|
+
schema_version: 1,
|
|
270
|
+
variant,
|
|
271
|
+
state: "not-selected",
|
|
272
|
+
evaluationType: group.type,
|
|
273
|
+
prompt: group.input.prompt ?? null,
|
|
274
|
+
prompts: group.input.turns ?? null,
|
|
275
|
+
activation: unavailableActivation(),
|
|
276
|
+
response: "",
|
|
277
|
+
transcript: [],
|
|
278
|
+
error: null,
|
|
279
|
+
};
|
|
280
|
+
}
|
|
281
|
+
|
|
282
|
+
function cancelledRun(group, variant) {
|
|
283
|
+
return {
|
|
284
|
+
schema_version: 1,
|
|
285
|
+
variant,
|
|
286
|
+
state: "cancelled",
|
|
287
|
+
evaluationType: group.type,
|
|
288
|
+
prompt: group.input.prompt ?? null,
|
|
289
|
+
prompts: group.input.turns ?? null,
|
|
290
|
+
activation: unavailableActivation(),
|
|
291
|
+
response: "",
|
|
292
|
+
transcript: [],
|
|
293
|
+
error: { kind: "cancellation", message: "Cancelled before subject execution" },
|
|
294
|
+
};
|
|
295
|
+
}
|
|
296
|
+
|
|
297
|
+
function invalidRun(group, variant, message) {
|
|
298
|
+
return {
|
|
299
|
+
schema_version: 1,
|
|
300
|
+
variant,
|
|
301
|
+
state: "invalid",
|
|
302
|
+
evaluationType: group.type,
|
|
303
|
+
prompt: group.input.prompt ?? null,
|
|
304
|
+
prompts: group.input.turns ?? null,
|
|
305
|
+
activation: unavailableActivation(),
|
|
306
|
+
response: "",
|
|
307
|
+
transcript: [],
|
|
308
|
+
error: { kind: "setup", message },
|
|
309
|
+
};
|
|
310
|
+
}
|
|
311
|
+
|
|
312
|
+
function persistenceFailedRun(run, error) {
|
|
313
|
+
return {
|
|
314
|
+
...run,
|
|
315
|
+
state: "infrastructure-failed",
|
|
316
|
+
error: {
|
|
317
|
+
kind: "persistence",
|
|
318
|
+
message: error instanceof Error ? error.message : String(error),
|
|
319
|
+
cause: run.error ?? null,
|
|
320
|
+
},
|
|
321
|
+
};
|
|
322
|
+
}
|
|
323
|
+
|
|
324
|
+
function infrastructureFailedRun(group, variant, error) {
|
|
325
|
+
return {
|
|
326
|
+
schema_version: 1,
|
|
327
|
+
variant,
|
|
328
|
+
state: "infrastructure-failed",
|
|
329
|
+
evaluationType: group.type,
|
|
330
|
+
prompt: group.input.prompt ?? null,
|
|
331
|
+
prompts: group.input.turns ?? null,
|
|
332
|
+
activation: unavailableActivation(),
|
|
333
|
+
response: "",
|
|
334
|
+
transcript: [],
|
|
335
|
+
error: { kind: "infrastructure", message: error instanceof Error ? error.message : String(error) },
|
|
336
|
+
};
|
|
337
|
+
}
|
|
338
|
+
|
|
339
|
+
function unavailableActivation() {
|
|
340
|
+
return {
|
|
341
|
+
targetRead: null,
|
|
342
|
+
firstReadTurn: null,
|
|
343
|
+
readTurns: [],
|
|
344
|
+
successfulReadPaths: [],
|
|
345
|
+
};
|
|
346
|
+
}
|