@tangle-network/agent-eval 0.122.5 → 0.122.6
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 +8 -0
- package/dist/benchmarks/index.js +2 -2
- package/dist/campaign/index.d.ts +22 -3
- package/dist/campaign/index.js +2 -2
- package/dist/{chunk-GLUBGL77.js → chunk-67H37Q6I.js} +77 -27
- package/dist/chunk-67H37Q6I.js.map +1 -0
- package/dist/{chunk-FLCXOTWF.js → chunk-YA6MJYZN.js} +11 -4
- package/dist/chunk-YA6MJYZN.js.map +1 -0
- package/dist/contract/index.d.ts +10 -0
- package/dist/contract/index.js +1 -1
- package/dist/index.js +2 -2
- package/dist/openapi.json +1 -1
- package/package.json +1 -1
- package/dist/chunk-FLCXOTWF.js.map +0 -1
- package/dist/chunk-GLUBGL77.js.map +0 -1
package/dist/contract/index.d.ts
CHANGED
|
@@ -1575,6 +1575,14 @@ interface ParetoParent {
|
|
|
1575
1575
|
label?: string;
|
|
1576
1576
|
rationale?: string;
|
|
1577
1577
|
}
|
|
1578
|
+
/** The lineage track that requested a proposal. */
|
|
1579
|
+
interface ProposalTrackContext {
|
|
1580
|
+
id: string;
|
|
1581
|
+
operation: 'extend' | 'branch' | 'merge';
|
|
1582
|
+
proposer: string;
|
|
1583
|
+
vision?: string;
|
|
1584
|
+
parentTrackIds: readonly string[];
|
|
1585
|
+
}
|
|
1578
1586
|
/** Exact measured state for the surface an optimizer is learning from.
|
|
1579
1587
|
* Unlike a model-authored expected gain, every value here comes from a
|
|
1580
1588
|
* completed campaign over the designed denominator. */
|
|
@@ -1629,6 +1637,8 @@ interface ProposeContext<TFindings = unknown> {
|
|
|
1629
1637
|
populationSize: number;
|
|
1630
1638
|
generation: number;
|
|
1631
1639
|
signal: AbortSignal;
|
|
1640
|
+
/** Present when a multi-track lineage requests this proposal. */
|
|
1641
|
+
track?: ProposalTrackContext;
|
|
1632
1642
|
/** Measured baseline for this optimization run. `runOptimization` always
|
|
1633
1643
|
* supplies it; optional for standalone proposer callers. */
|
|
1634
1644
|
baselineOutcome?: ScoredSurfaceOutcome;
|
package/dist/contract/index.js
CHANGED
package/dist/index.js
CHANGED
|
@@ -53,7 +53,7 @@ import {
|
|
|
53
53
|
pairArms,
|
|
54
54
|
parseCorrectnessResponse,
|
|
55
55
|
verifyCompletion
|
|
56
|
-
} from "./chunk-
|
|
56
|
+
} from "./chunk-67H37Q6I.js";
|
|
57
57
|
import {
|
|
58
58
|
DEFAULT_MUTATION_PRIMITIVES,
|
|
59
59
|
DEFAULT_RED_TEAM_CORPUS,
|
|
@@ -91,7 +91,7 @@ import {
|
|
|
91
91
|
scoreRedTeamOutput,
|
|
92
92
|
surfaceContentHash,
|
|
93
93
|
toolNamesForRun
|
|
94
|
-
} from "./chunk-
|
|
94
|
+
} from "./chunk-YA6MJYZN.js";
|
|
95
95
|
import {
|
|
96
96
|
BackendIntegrityError,
|
|
97
97
|
assertRealAgentReceipts,
|
package/dist/openapi.json
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
"openapi": "3.1.0",
|
|
3
3
|
"info": {
|
|
4
4
|
"title": "@tangle-network/agent-eval — wire protocol",
|
|
5
|
-
"version": "0.122.
|
|
5
|
+
"version": "0.122.6",
|
|
6
6
|
"description": "HTTP and stdio RPC interface to agent-eval. The TypeScript runtime is the source of truth; this spec is the contract that cross-language clients (Python, Rust, Go) generate from.\n\nWire-protocol version: 1.0.0. Bumps on breaking changes to request/response schemas.",
|
|
7
7
|
"contact": {
|
|
8
8
|
"name": "Tangle Network",
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@tangle-network/agent-eval",
|
|
3
|
-
"version": "0.122.
|
|
3
|
+
"version": "0.122.6",
|
|
4
4
|
"description": "Evaluate and improve AI agents from runs, traces, judges, and feedback. Compare candidates, cluster failures, measure lift, and gate releases.",
|
|
5
5
|
"homepage": "https://github.com/tangle-network/agent-eval#readme",
|
|
6
6
|
"repository": {
|