Orkas is an open-source (MIT), local-first AI team for your desktop: a Commander turns your goal into to-dos and coordinates 16 built-in specialist agents plus Claude Code, Codex, OpenCode, OpenClaw and Hermes, with projects, scheduled automations and connectors. Free app for macOS & Windows; built-in models or your own keys.
JavaScript agent-collaborationagent-harnessagent-orchestrationagent-teamagent-workflowai-assistantai-automationai-employeeai-productivityai-teamai-workflowautonomous-agentsclaude-code-guiclaude-coworkdesktop-agentlocal-firstmulti-agentmulti-agent-systemself-evolving-agentstask-automation
Use no-op debug recorder in OrkasOpen
wytd39743bc75ba
diff --git a/src/main/model/core-agent/client.ts b/src/main/model/core-agent/client.tsindex 327d6942..490b13d1 100644--- a/src/main/model/core-agent/client.ts+++ b/src/main/model/core-agent/client.ts@@ -42,8 +42,14 @@ interface DebugRecorder {finish(result: { text: string; aborted: boolean; error: string | null }): void;}-function createDebugRecorder(): DebugRecorder | null {- return null;+const NOOP_DEBUG_RECORDER: DebugRecorder = {+ record() {},+ setActiveCandidate() {},+ finish() {},+};++function createDebugRecorder(): DebugRecorder {+ return NOOP_DEBUG_RECORDER;}/**@@ -244,10 +250,9 @@ export async function* streamChatWithModel(opts: ChatOptions): AsyncGenerator<Stelse abortSignal.addEventListener('abort', onExternalAbort, { once: true });}- // Dev-only archiver — no-op in prod. Initialised after buildRunner so- // the record can carry the resolved provider / model ids for prompt- // triage ("did this call actually hit the model I think it did").- let recorder: DebugRecorder | null = null;+ // PC dev builds attach an archive recorder here. OrkasOpen keeps the same+ // interface as a no-op object so stripped debug code cannot affect model calls.+ const recorder = createDebugRecorder();let finalText = '';let errText: string | null = null;let abortedFlag = false;@@ -280,7 +285,7 @@ export async function* streamChatWithModel(opts: ChatOptions): AsyncGenerator<St...(onSkillAdvertised ? { onSkillAdvertised } : {}),...(onSkillInvoked ? { onSkillInvoked } : {}),onNativeSearchInjected: (info) => {- recorder?.record({+ recorder.record({type: 'progress',event: { stream: 'native_search', data: { phase: 'injected', ...info } },});@@ -293,13 +298,11 @@ export async function* streamChatWithModel(opts: ChatOptions): AsyncGenerator<St// when complete()/stream() actually picks a candidate, so the recorder// is always live by then.onCandidateChosen: (info) => {- recorder?.setActiveCandidate(info);+ recorder.setActiveCandidate(info);},});const { runner, providerId, modelId, resolvedSystemPrompt, profileId, entryId, toolDefs, skillDisplayNameById, agentDisplayNameById } = built;- recorder = createDebugRecorder();-const sandboxEnv = buildSkillSandboxEnv();log.info(`turn start ${turnTag} user=${userId} provider=${providerId} model=${modelId}`);@@ -329,7 +332,7 @@ export async function* streamChatWithModel(opts: ChatOptions): AsyncGenerator<Stfor await (const ev of mapCoreAgentEvents(captureResult(rawEvents), { userId, skillDisplayNameById, agentDisplayNameById })) {resetIdle();eventCount += 1;- recorder?.record(ev as any);+ recorder.record(ev as any);if (ev.type === 'final') finalText = (ev as any).text || finalText;if (ev.type === 'error') { errText = (ev as any).text || errText; if ((ev as any).aborted) abortedFlag = true; }// NOTE: compaction summaries are deliberately NOT mined into cross-session@@ -392,10 +395,8 @@ export async function* streamChatWithModel(opts: ChatOptions): AsyncGenerator<St}releaseSlotOnce('finally');releaseSessionOnce('finally');- if (recorder) {- try { recorder.finish({ text: finalText, aborted: abortedFlag, error: errText }); }- catch (err) { log.warn('archive finish failed:', err); }- }+ try { recorder.finish({ text: finalText, aborted: abortedFlag, error: errText }); }+ catch (err) { log.warn('archive finish failed:', err); }yield { type: 'done' };}}