@agentv/sdk 4.41.3-next.1

package/README.md

@@ -0,0 +1,71 @@
+# @agentv/sdk
+
+Evaluation SDK for AgentV - build custom graders and prompt templates around the canonical AgentV eval model.
+
+## Installation
+
+```bash
+npm install @agentv/sdk
+```
+
+## Quick Start
+
+### defineAssertion (simplest way)
+
+```typescript
+#!/usr/bin/env bun
+import { defineAssertion } from '@agentv/sdk';
+
+export default defineAssertion(({ output }) => ({
+  pass: (output ?? '').toLowerCase().includes('hello'),
+  reasoning: 'Checks for greeting',
+}));
+```
+
+Assertions support `pass: boolean` for simple checks and `score: number` (0-1) for granular scoring.
+
+### defineCodeGrader (full control)
+
+```typescript
+#!/usr/bin/env bun
+import { defineCodeGrader } from '@agentv/sdk';
+
+export default defineCodeGrader(({ output, traceSummary }) => ({
+  score: (output ?? '').length > 0 ? 1.0 : 0.0,
+  assertions: [
+    { text: 'Output received', passed: (output ?? '').length > 0 },
+    { text: 'Trace summary available', passed: traceSummary !== null },
+  ],
+}));
+```
+
+Both functions handle stdin/stdout parsing, snake_case conversion, Zod validation, and error handling automatically.
+
+## Exports
+
+- `defineAssertion(handler)` - Define a custom assertion (pass/fail + optional score)
+- `defineCodeGrader(handler)` - Define a code grader grader (full score control)
+- `definePromptTemplate(handler)` - Define a dynamic prompt template
+- `AssertionContext`, `AssertionScore` - Assertion types
+- `CodeGraderInput`, `CodeGraderResult` - Code grader types
+- `TraceSummary`, `Message`, `ToolCall` - Trace data types
+- `createTargetClient()` - LLM target proxy for graders
+- `z` - Re-exported Zod for custom config schemas
+
+## Documentation
+
+For complete documentation including:
+- Full input/output schemas
+- Typed config examples
+- Execution metrics usage
+- Best practices
+
+See the docs site guides under `apps/web/src/content/docs/docs/graders/` or run `agentv skills get agentv-eval-writer`.
+
+## Repository
+
+[https://github.com/EntityProcess/agentv](https://github.com/EntityProcess/agentv)
+
+## License
+
+MIT License - see [LICENSE](../../LICENSE) for details.

package/package.json

@@ -0,0 +1,37 @@
+{
+  "name": "@agentv/sdk",
+  "version": "4.41.3-next.1",
+  "description": "Evaluation SDK for AgentV - build custom code judges",
+  "type": "module",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/EntityProcess/agentv.git"
+  },
+  "homepage": "https://agentv.dev",
+  "bugs": {
+    "url": "https://github.com/EntityProcess/agentv/issues"
+  },
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "require": "./dist/index.cjs"
+    }
+  },
+  "scripts": {
+    "prepublishOnly": "node -e \"if(process.env.ALLOW_PUBLISH!=='1'){console.error('ERROR: Use bun run publish:next, then bun run promote:latest');process.exit(1)}\"",
+    "build": "tsup",
+    "dev": "tsup --watch",
+    "typecheck": "tsc --noEmit",
+    "lint": "biome check .",
+    "format": "biome format --write .",
+    "fix": "biome check --write .",
+    "test": "bun test"
+  },
+  "files": ["dist", "README.md"],
+  "dependencies": {
+    "zod": "^3.23.8"
+  }
+}