@wix/evalforge-evaluator 0.105.0 → 0.107.0

package/build/types/api-client.d.ts

@@ -1,10 +1,9 @@
 /**
  * API Client for fetching data from the eval server.
  */
-import type { EvalRun, EvalRunResult, SkillsGroup, TestScenario, Template, SkillWithLatestVersion, SkillVersion, Agent, CustomAssertion, MCPEntity, SubAgent, Rule } from '@wix/evalforge-types';
+import type { EvalRun, EvalRunResult, TestScenario, Template, SkillWithLatestVersion, SkillVersion, Agent, Preset, CustomAssertion, MCPEntity, SubAgent, Rule } from '@wix/evalforge-types';
 export interface ApiClient {
     getEvalRun(projectId: string, id: string): Promise<EvalRun>;
-    getSkillsGroup(projectId: string, id: string): Promise<SkillsGroup>;
     getScenario(projectId: string, id: string): Promise<TestScenario>;
     getSkill(projectId: string, id: string): Promise<SkillWithLatestVersion>;
     getSkillVersion(projectId: string, skillId: string, versionId: string): Promise<SkillVersion>;
@@ -14,6 +13,7 @@ export interface ApiClient {
     getMcp(projectId: string, id: string): Promise<MCPEntity>;
     getSubAgent(projectId: string, id: string): Promise<SubAgent>;
     getRule(projectId: string, id: string): Promise<Rule>;
+    getPreset(projectId: string, id: string): Promise<Preset>;
     getAssertion(projectId: string, id: string): Promise<CustomAssertion>;
     addResult(projectId: string, evalRunId: string, result: EvalRunResult): Promise<void>;
     clearResults(projectId: string, evalRunId: string): Promise<void>;

package/build/types/fetch-evaluation-data.d.ts

@@ -1,4 +1,4 @@
-import type { EvalRun, TestScenario, SkillWithLatestVersion, SkillsGroup, Agent, Template, CustomAssertion, Assertion, MCPEntity, SubAgent, Rule } from '@wix/evalforge-types';
+import type { EvalRun, TestScenario, SkillWithLatestVersion, Agent, Template, CustomAssertion, Assertion, MCPEntity, SubAgent, Rule } from '@wix/evalforge-types';
 import { type SystemAssertionId } from '@wix/evalforge-types';
 import type { ApiClient } from './api-client.js';
 /**
@@ -31,15 +31,14 @@ export interface ScenarioItem {
 }
 /**
  * Data shape returned by fetchEvaluationData for the run flow.
- * Run-level agent + skills + skills group + mcps + subAgents; per-scenario template.
+ * Run-level agent + skills + mcps + subAgents; per-scenario template.
  */
 export interface EvaluationData {
     evalRun: EvalRun;
     agent: Agent | null;
     skills: SkillWithLatestVersion[];
-    skillsGroup: SkillsGroup | null;
-    /** Display name for the skills group (from skillsGroup.name when present) */
-    skillsGroupName: string;
+    /** Display name: preset name from DB when available, otherwise joined skill names */
+    presetName: string;
     mcps: MCPEntity[];
     subAgents: SubAgent[];
     rules: Rule[];
@@ -48,8 +47,8 @@ export interface EvaluationData {
 /**
  * Fetch all data needed to run an evaluation.
  *
- * Uses evalRun.agentId and evalRun.skillsGroupId as run-level inputs.
- * Loads skills from the skills group; loads code agent; loads scenarios and templates by scenario.templateId.
+ * Uses evalRun.agentId and evalRun.skillIds as run-level inputs.
+ * Loads skills by ID; loads code agent; loads scenarios and templates by scenario.templateId.
  *
  * @throws Error if eval run has no code agent when expected
  * @throws Error if eval run has no skills group or skills group has no skills when expected

package/build/types/run-scenario/agents/claude-code/write-sub-agents.d.ts

@@ -1,12 +1,16 @@
-import type { SubAgent } from '@wix/evalforge-types';
+import type { SubAgent, GitHubSource } from '@wix/evalforge-types';
+import { type FetchGitHubFolderOptions } from '@wix/evalforge-github-client';
+/** Signature for the single-file GitHub fetch function (injectable for tests). */
+export type FetchGitHubFileFn = (source: GitHubSource, options?: FetchGitHubFolderOptions) => Promise<string>;
 /**
  * Write sub-agent markdown files to .claude/agents/ for Claude Code to discover.
  *
- * Each sub-agent's subAgentMd (YAML frontmatter + body) is written to
- * .claude/agents/{name}.md. Claude Code loads these at session start.
+ * For sub-agents with a GitHub source, the latest .md file is live-fetched.
+ * For inline sub-agents, the stored subAgentMd is used directly.
  *
  * @see https://code.claude.com/docs/en/sub-agents#write-subagent-files
  * @param cwd - Working directory (project root for Claude Code)
  * @param subAgents - Sub-agent entities to write
+ * @param fetchFn - Optional fetch function for testing (defaults to fetchGitHubFile)
  */
-export declare function writeSubAgentsToFilesystem(cwd: string, subAgents: SubAgent[]): Promise<void>;
+export declare function writeSubAgentsToFilesystem(cwd: string, subAgents: SubAgent[], fetchFn?: FetchGitHubFileFn): Promise<void>;

package/build/types/run-scenario/index.d.ts

@@ -10,7 +10,7 @@ export type { ScenarioItem, EvaluationData } from '../fetch-evaluation-data.js';
  * @param config - Evaluator configuration
  * @param evalRunId - The evaluation run ID
  * @param scenario - The test scenario to run
- * @param evalData - Fetched evaluation data (skills, skillsGroup, agent, mcps, subAgents)
+ * @param evalData - Fetched evaluation data (skills, agent, mcps, subAgents)
  * @param template - Optional pre-fetched template entity
  * @param resolvedAssertions - Optional assertions resolved from assertionIds
  * @returns Complete evaluation result

package/build/types/run-scenario/run-agent-with-context.d.ts

@@ -12,7 +12,7 @@ import type { EvaluationData } from '../fetch-evaluation-data.js';
  * @param config - Evaluator configuration
  * @param evalRunId - The evaluation run ID (for live trace context)
  * @param scenario - The test scenario to run
- * @param evalData - Fetched evaluation data (skills, skillsGroup, agent, mcps, subAgents)
+ * @param evalData - Fetched evaluation data (skills, agent, mcps, subAgents)
  * @param workDir - Optional working directory for the scenario
  * @returns Partial result without assertion fields
  */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@wix/evalforge-evaluator",
-  "version": "0.105.0",
+  "version": "0.107.0",
   "description": "EvalForge Evaluator",
   "bin": "./build/index.js",
   "files": [
@@ -22,8 +22,8 @@
     "@anthropic-ai/claude-agent-sdk": "^0.2.49",
     "@anthropic-ai/claude-code": "^2.1.49",
     "@wix/eval-assertions": "0.23.0",
-    "@wix/evalforge-github-client": "0.23.0",
-    "@wix/evalforge-types": "0.48.0",
+    "@wix/evalforge-github-client": "0.25.0",
+    "@wix/evalforge-types": "0.50.0",
     "ai": "^6.0.93",
     "diff": "^7.0.0",
     "tar": "^7.5.3",
@@ -62,5 +62,5 @@
       "artifactId": "evalforge-evaluator"
     }
   },
-  "falconPackageHash": "7f3f2d1b27312daf480e195d122e0885ac476d0f881100f407a34755"
+  "falconPackageHash": "59ea86dd554511d348988564544eee3ea330a005592d9aa9a7d22719"
 }