daemon.cjs · part 368

60 text occurrences from desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/agent-host-daemon/dist/bin/daemon.cjs, part 368. Every entry preserves the shipped literal and its saved verdict or selection reason.

File contents and all parts · All files

use strict

use strict

Failed to record private inference usage

Private inference credential helper failed

Private inference gateway rejected the credential (HTTP ${current.statusCode})

Inference gateway rejected the credential

The private inference gateway${host === void 0 ? "" : ` at ${host}`} returned HTTP ${current.statusCode}${excerpt === void 0 ? "" : `: ${excerpt}`}

 at ${host}

use strict

Re-run your credential helper's sign-in step (for example your SSO login), then send the message again.

Provider metadata unavailable

Local agent stream error

Local agent stream error

use strict

${baseUrl} ${modelId}

use strict

[image omitted: this gateway accepts text tool results only]

[image omitted: this model is not configured for image input; set vision = true on its private-inference.toml entry if it accepts images]

use strict

Private inference config: no configuration present

use strict

Could not reach your private inference gateway at ${baseUrl}.

Could not reach your private inference gateway at ${state.baseUrl}. Check your network or VPN, then try again.

Private inference gateway unreachable

Private inference gateway unreachable

use strict

Bearer ${options2.apiKey}

Search the web for the following query and return a concise answer with sources:
${args.searchTerm}

Provider request failed: ${response.status} ${response.statusText}
${responseText}

use strict

${tool} is disabled in private inference: it would send content to Cursor's backend.

Web search

Web fetch

Image generation

use strict

Private inference is enabled but has no endpoint configuration. Put private-inference.toml in the enterprise Cursor directory (a Grok Build config.toml: [model_providers.<name>] with base_url, api_backend = "responses" and env_key or [auth] auth_provider_command — an executable path that prints the bare token — one [model.<key>] per model, [models] default), or for development set CURSOR_PRIVATE_INFERENCE_CONFIG=<config.toml> or the CURSOR_PRIVATE_INFERENCE_BASE_URL / _MODEL shorthand, then restart, or ask your admin to disable private inference for your team.

Private inference requires durable usage storage and an authenticated usage-reporting transport.

## Model requests (private inference)

${runs.length} run${runs.length === 1 ? "" : "s"} captured on this device. Each request below is what the client sent to the configured gateway and what the gateway returned; nothing here was sent to Cursor.

### Run ${runIndex + 1} — ${run.generationUUID}

- Status: ${run.terminalStatus}

- Session: ${run.conversationId}

- Captured: ${run.createdAt}

- Requests: ${run.invocations.length}

${lines2.join("\n").trimEnd()}

#### Request ${index} — ${invocation.modelId}

- Provider: ${invocation.providerName}

- Status: ${invocation.status}${invocation.errorMessage === void 0 ? "" : ` (${invocation.errorMessage})`}

 (${invocation.errorMessage})

- Attempt: ${invocation.attempt} (${invocation.attemptRequestId})

- Started: ${invocation.startedAt}; completed: ${invocation.completedAt}

- Tokens: ${invocation.promptTokens} prompt + ${invocation.completionTokens} completion = ${invocation.totalTokens}${invocation.tokenLimit === void 0 ? "" : ` (limit ${invocation.tokenLimit})`}

 (limit ${invocation.tokenLimit})

- Tools offered (${invocation.tools.length}): ${invocation.tools.map((tool) => `\`${tool.name}\``).join(", ") || "none"}

_No response messages._

> **(undecodable message)**

_reasoning (redacted)_

_image${part.mimeType === void 0 ? "" : ` (${part.mimeType})`}_

 (${part.mimeType})

_file (${part.mimeType})_