@wix/evalforge-evaluator 0.176.0 → 0.178.0

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/README.md CHANGED
@@ -35,6 +35,8 @@ evaluator <project-id> <eval-run-id>
35
35
  | `EVAL_ROUTE_HEADER` | No | `x-wix-route` header for deploy preview routing |
36
36
  | `EVAL_AUTH_TOKEN` | No | Bearer token for public endpoint authentication |
37
37
 
38
+ For OpenCode runs, the evaluator sets `lsp: false` in `OPENCODE_CONFIG_CONTENT` and `OPENCODE_DISABLE_LSP_DOWNLOAD` / `OPENCODE_DISABLE_FILETIME_CHECK` in the process environment (same as ditto `codegen`) to avoid LSP hangs after edit tools and spurious "file modified since last read" failures in automated evals.
39
+
38
40
  The evaluator is typically launched by the backend (locally or on a remote Dev Machine) with these variables pre-configured.
39
41
 
40
42
  ## Scripts
package/build/index.js CHANGED
@@ -2520,6 +2520,8 @@ async function buildOpenCodeEnv(options) {
2520
2520
  provider,
2521
2521
  ...Object.keys(agentOverrides).length > 0 ? { agent: { build: agentOverrides } } : {},
2522
2522
  permission,
2523
+ // Align with ditto/codegen: LSP post-edit diagnostics can hang remote eval machines.
2524
+ lsp: false,
2523
2525
  ...mcp ? { mcp } : {},
2524
2526
  ...instructions ? { instructions } : {}
2525
2527
  };
@@ -2527,7 +2529,8 @@ async function buildOpenCodeEnv(options) {
2527
2529
  ...process.env,
2528
2530
  PATH: ensureOpenCodeInPath(process.env.PATH || ""),
2529
2531
  OPENCODE_CONFIG_CONTENT: JSON.stringify(config),
2530
- OPENCODE_DISABLE_LSP_DOWNLOAD: "true"
2532
+ OPENCODE_DISABLE_LSP_DOWNLOAD: "true",
2533
+ OPENCODE_DISABLE_FILETIME_CHECK: "true"
2531
2534
  };
2532
2535
  return { env, providerID, modelID };
2533
2536
  }
@@ -3687,7 +3690,8 @@ var MODEL_PRICING = {
3687
3690
  // Google Gemini 3.x — standard pricing up to 200K context tokens
3688
3691
  "gemini-3-pro-preview": { input: 2, output: 12 },
3689
3692
  "gemini-3-flash-preview": { input: 0.5, output: 3 },
3690
- "gemini-3.1-pro-preview": { input: 2, output: 12 }
3693
+ "gemini-3.1-pro-preview": { input: 2, output: 12 },
3694
+ "gemini-3.5-flash": { input: 1.5, output: 9 }
3691
3695
  };
3692
3696
  function extractGatewayCost(step, provider) {
3693
3697
  try {