e81f71bdac
* feat(evals): add ExecutionBalanceEvaluator, docs, tests, version bumps * chore: remove obsolete package-lock in evals/framework pre-rebase * chore: translate ExecutionBalanceEvaluator and test files from Spanish to English - Translate all code comments and documentation in execution-balance-evaluator.ts - Translate test descriptions and prompts in execution-balance-positive.yaml - Translate test descriptions and prompts in execution-balance-negative.yaml This ensures consistency with the rest of the English codebase as requested in PR review. * fix(docs): translate Spanish content to English in documentation --------- Co-authored-by: Alexander Daza <dev.alexander@example.com> Co-authored-by: Darren Hinde <107584450+darrenhinde@users.noreply.github.com>
66 lines
4.1 KiB
JSON
66 lines
4.1 KiB
JSON
{
|
|
"name": "opencode-agents",
|
|
"version": "0.0.2",
|
|
"description": "OpenCode agent evaluation framework and test suites",
|
|
"private": true,
|
|
"workspaces": [
|
|
"evals/framework"
|
|
],
|
|
"scripts": {
|
|
"test": "npm run test:all",
|
|
"test:all": "cd evals/framework && npm run eval:sdk",
|
|
"test:core": "npm run test:openagent:core",
|
|
"test:openagent": "cd evals/framework && npm run eval:sdk -- --agent=openagent",
|
|
"test:openagent:core": "cd evals/framework && npm run eval:sdk:core -- --agent=openagent",
|
|
"test:opencoder": "cd evals/framework && npm run eval:sdk -- --agent=opencoder",
|
|
"test:openagent:grok": "npm run test:openagent -- --model=opencode/grok-code-fast",
|
|
"test:openagent:claude": "npm run test:openagent -- --model=anthropic/claude-3-5-sonnet-20241022",
|
|
"test:openagent:gpt4": "npm run test:openagent -- --model=openai/gpt-4-turbo",
|
|
"test:opencoder:grok": "npm run test:opencoder -- --model=opencode/grok-code-fast",
|
|
"test:opencoder:claude": "npm run test:opencoder -- --model=anthropic/claude-3-5-sonnet-20241022",
|
|
"test:opencoder:gpt4": "npm run test:opencoder -- --model=openai/gpt-4-turbo",
|
|
"test:all:grok": "npm run test:all -- --model=opencode/grok-code-fast",
|
|
"test:all:claude": "npm run test:all -- --model=anthropic/claude-3-5-sonnet-20241022",
|
|
"test:all:gpt4": "npm run test:all -- --model=openai/gpt-4-turbo",
|
|
"test:pattern": "cd evals/framework && npm run eval:sdk -- --pattern",
|
|
"test:debug": "cd evals/framework && npm run eval:sdk -- --debug",
|
|
"test:openagent:developer": "npm run test:openagent -- --pattern='developer/*.yaml'",
|
|
"test:openagent:context": "npm run test:openagent -- --pattern='context-loading/*.yaml'",
|
|
"test:openagent:business": "npm run test:openagent -- --pattern='business/*.yaml'",
|
|
"test:opencoder:developer": "npm run test:opencoder -- --pattern='developer/*.yaml'",
|
|
"test:opencoder:bash": "npm run test:opencoder -- --pattern='developer/bash-*.yaml'",
|
|
"test:ci": "npm run test:ci:openagent && npm run test:ci:opencoder",
|
|
"test:ci:openagent": "npm run test:openagent -- --pattern='smoke-test.yaml' --no-evaluators",
|
|
"test:ci:opencoder": "npm run test:opencoder -- --pattern='developer/simple-bash-test.yaml' --no-evaluators",
|
|
"dashboard": "cd evals/results && ./serve.sh",
|
|
"dashboard:open": "npm run dashboard && open http://localhost:8000",
|
|
"results:openagent": "echo 'OpenAgent results:' && ls -lh evals/results/history/*openagent*.json 2>/dev/null | tail -5 || echo 'No results yet'",
|
|
"results:opencoder": "echo 'OpenCoder results:' && ls -lh evals/results/history/*opencoder*.json 2>/dev/null | tail -5 || echo 'No results yet'",
|
|
"results:latest": "cat evals/results/latest.json 2>/dev/null | jq '.agent, .passed, .failed' || echo 'No results yet'",
|
|
"version": "node -p \"require('./package.json').version\"",
|
|
"version:bump": "./scripts/versioning/bump-version.sh",
|
|
"version:bump:patch": "npm version patch --no-git-tag-version && node -p \"require('./package.json').version\" > VERSION",
|
|
"version:bump:minor": "npm version minor --no-git-tag-version && node -p \"require('./package.json').version\" > VERSION",
|
|
"version:bump:major": "npm version major --no-git-tag-version && node -p \"require('./package.json').version\" > VERSION",
|
|
"version:bump:alpha": "npm version prerelease --preid=alpha --no-git-tag-version && node -p \"require('./package.json').version\" > VERSION",
|
|
"version:bump:beta": "npm version prerelease --preid=beta --no-git-tag-version && node -p \"require('./package.json').version\" > VERSION",
|
|
"version:bump:rc": "npm version prerelease --preid=rc --no-git-tag-version && node -p \"require('./package.json').version\" > VERSION",
|
|
"dev:setup": "cd evals/framework && npm install",
|
|
"dev:build": "cd evals/framework && npm run build",
|
|
"dev:test": "cd evals/framework && npm test",
|
|
"dev:clean": "cd evals/framework && rm -rf dist node_modules && npm install"
|
|
},
|
|
"keywords": [
|
|
"opencode",
|
|
"agents",
|
|
"evaluation",
|
|
"testing",
|
|
"ai"
|
|
],
|
|
"author": "Darren Hinde",
|
|
"license": "MIT",
|
|
"repository": {
|
|
"type": "git",
|
|
"url": "https://github.com/darrenhinde/OpenAgents.git"
|
|
}
|
|
}
|