@wix/evalforge-types 0.41.0 → 0.42.0
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/build/index.js +17 -2
- package/build/index.js.map +2 -2
- package/build/index.mjs +16 -2
- package/build/index.mjs.map +2 -2
- package/build/types/agent/adapter.d.ts +7 -0
- package/build/types/index.d.ts +1 -1
- package/build/types/target/agent.d.ts +10 -1
- package/package.json +2 -2
package/build/index.mjs
CHANGED
|
@@ -994,6 +994,15 @@ var TargetSchema = TenantEntitySchema.extend({
|
|
|
994
994
|
|
|
995
995
|
// src/target/agent.ts
|
|
996
996
|
import { z as z6 } from "zod";
|
|
997
|
+
var DEFAULT_EVALUATOR_SYSTEM_PROMPT = `IMPORTANT: This is an automated evaluation run. Follow these guidelines:
|
|
998
|
+
1. Execute the requested changes immediately without asking for confirmation.
|
|
999
|
+
2. Do NOT ask "would you like me to proceed?" or similar questions.
|
|
1000
|
+
3. Do NOT use the Task tool to delegate simple operations - do them directly yourself.
|
|
1001
|
+
4. Keep your approach simple and direct - avoid excessive planning.
|
|
1002
|
+
5. Make targeted edits using Read and Edit tools rather than exploring the entire codebase.
|
|
1003
|
+
6. If you encounter an error, fix it directly rather than starting over.
|
|
1004
|
+
7. Your project root is the current working directory. Always create and modify source code files relative to the project root, NOT inside .claude/skills/ directories.
|
|
1005
|
+
8. Before finishing, run the project's package manager install command (e.g. \`npm install\`, \`yarn install\`, or \`pnpm install\` depending on the lockfile present) to ensure all dependencies are installed and the project is ready to build.`;
|
|
997
1006
|
var AgentRunCommand = /* @__PURE__ */ ((AgentRunCommand2) => {
|
|
998
1007
|
AgentRunCommand2["CLAUDE"] = "claude";
|
|
999
1008
|
return AgentRunCommand2;
|
|
@@ -1007,7 +1016,10 @@ var AgentSchema = TargetSchema.extend({
|
|
|
1007
1016
|
/** Command to run the agent */
|
|
1008
1017
|
runCommand: AgentRunCommandSchema,
|
|
1009
1018
|
/** Optional model configuration override */
|
|
1010
|
-
modelConfig: ModelConfigSchema.optional()
|
|
1019
|
+
modelConfig: ModelConfigSchema.optional(),
|
|
1020
|
+
systemPrompt: z6.string().nullish().describe(
|
|
1021
|
+
"Override for eval runs. undefined=default instructions, null=raw agent, string=append to claude_code preset. See https://docs.anthropic.com/en/docs/claude-code/sdk/modifying-system-prompts"
|
|
1022
|
+
)
|
|
1011
1023
|
});
|
|
1012
1024
|
var CreateAgentInputSchema = AgentSchema.omit({
|
|
1013
1025
|
id: true,
|
|
@@ -1016,7 +1028,8 @@ var CreateAgentInputSchema = AgentSchema.omit({
|
|
|
1016
1028
|
deleted: true
|
|
1017
1029
|
});
|
|
1018
1030
|
var UpdateAgentInputSchema = CreateAgentInputSchema.partial().extend({
|
|
1019
|
-
modelConfig: ModelConfigSchema.optional().nullable()
|
|
1031
|
+
modelConfig: ModelConfigSchema.optional().nullable(),
|
|
1032
|
+
systemPrompt: z6.string().optional().nullable()
|
|
1020
1033
|
});
|
|
1021
1034
|
|
|
1022
1035
|
// src/target/skill.ts
|
|
@@ -2222,6 +2235,7 @@ export {
|
|
|
2222
2235
|
CreateTestScenarioInputSchema,
|
|
2223
2236
|
CreateTestSuiteInputSchema,
|
|
2224
2237
|
CustomAssertionSchema,
|
|
2238
|
+
DEFAULT_EVALUATOR_SYSTEM_PROMPT,
|
|
2225
2239
|
DEFAULT_JUDGE_MODEL,
|
|
2226
2240
|
DiffContentSchema,
|
|
2227
2241
|
DiffLineSchema,
|