107
distinct LLM tasks found
across 177 call sites · 0 routed through Atlas
Repositories scanned146/147
30 contain an LLM call · 4 have no commits, so there was nothing to read · 1 not yet reached
No app routes through Atlas yet, so every number here is read from source, not measured at runtime. Point one app's traffic at the gateway and its real calls, tokens and cost appear.
224 of 224 call sites · 137 tasks · 28 apps
Where the calls go
Click a bar to filter. Bars always show the whole estate at one shared scale, so any bar compares to any other — a selection is shown by the bar being pressed, and the lighter portion is how much of it the current filter leaves visible.
Provider
Kind
Tasks
Each row is one task narrowed to the calls that match — so the counts here always sum to the filter above.
AppTaskModelKindCallsRouting
UnitreeTts speech synthesistts-speech-synthesiseleven_turbo_v2_5audio8not routed
Unitreevoice/bob.py:1017
Unitreevoice/build_augmented_dataset.py:83
Unitreevoice/speak.py:127
Unitreevoice/speak.py:279
Unitreevoice/speak.py:339
Unitreevoice/speak.py:459
Unitreevoice/speak.py:553
Unitreevoice/speak.py:736
BloomImage scene generationimage-scene-generationfal-ai/flux/dev, fal-ai/flux/dev/image-to-imageimage7not routed
Bloomapp/api/components/mural/route.js:59
Bloomapp/api/logo/livestage/route.js:61
Bloomapp/api/logo/stages/route.js:99
Bloomlib/workflow.js:103
Bloomlib/workflow.js:107
Bloomscripts/create-asset.mjs:100
Bloomscripts/create-logo.mjs:86
peoplesoft_analysisunlinkedEmbed test questionquestion-embeddingchosen at runtimeembedding6not routed
Embeds an admin's sample question so it can be run through table retrieval to preview what a permission set would allow.
peoplesoft_analysisdashboard/routes/permissions.py:256
peoplesoft_analysispipeline/chatbot.py:925
peoplesoft_analysisscripts/qa_f8e.py:65
peoplesoft_analysisscripts/qa_f8f.py:108
peoplesoft_analysisscripts/qa_f8g.py:110
peoplesoft_analysisscripts/test_embed_retrieval.py:118
BloomImage to video animationimage-to-video-animationfal-ai/kling-video/v2.1/standard/image-to-videovideo6not routed
Bloomapp/api/components/mural/route.js:107
Bloomapp/api/logo/livestage/route.js:104
Bloomapp/api/logo/stages/route.js:145
Bloomlib/workflow.js:208
Bloomscripts/create-asset.mjs:119
Bloomscripts/create-logo.mjs:105
UnitreeRealtime voice sessionrealtime-voice-sessiongpt-4o-realtime-preview, gpt-realtime-2audio6not routed
Unitreebob-dashboard/src/app/(g1)/talk/use-realtime.ts:587
Unitreebob-dashboard/src/app/api/realtime-session/route.ts:40
Unitreevoice/bob.py:589
Unitreevoice/live.py:151
Unitreevoice/openai_voice.py:77
Unitreevoice/realtime.py:125
UnitreeSpeech transcriptionspeech-transcriptiongpt-4o-mini-transcribe, gpt-realtime, whisper-1audio6not routed
Unitreebob-dashboard/src/app/api/transcribe/route.ts:51
Unitreevoice/dashboard_listener.py:482
Unitreevoice/dashboard_listener.py:632
Unitreevoice/listen.py:57
Unitreevoice/listen.py:97
Unitreevoice/wakeword.py:163
CortexToken countingtoken-countingclaude-haiku-4-5-20251001token counting6not routed
Cortexcore/cortex/apollo/proxy/client.py:488
Cortexcore/cortex/praxis/api/routes/ai_keys.py:289
Cortexcore/cortex/praxis/api/routes/chat.py:2518
Cortexcore/cortex/praxis/api/routes/chat.py:2528
Cortexcore/cortex/praxis/api/routes/chat.py:2535
Cortexcore/cortex/praxis/api/routes/chat.py:2542
UnitreeConversational responseconversational-responsechosen at runtimetext4not routed
Unitreevoice/brain.py:259
Unitreevoice/brain.py:310
Unitreevoice/brain.py:344
Unitreevoice/brain.py:445
peoplesoft_analysisunlinkedEmbed column semantic textschema-embedding-ingestchosen at runtimeembedding4not routed
Batch-embeds generated semantic descriptions of table columns for storage in the column_embeddings table.
peoplesoft_analysispipeline/ingest_column_embeddings.py:463
peoplesoft_analysispipeline/ingest_embeddings.py:205
peoplesoft_analysispipeline/ingest_process_embeddings.py:171
peoplesoft_analysispipeline/ingest_sql_embeddings.py:289
athena-v2Commit review classificationcommit-review-classificationchosen at runtimetext3not routed
Athenaextensions/sonance-cortex/index.ts:2458
Athenaextensions/sonance-cortex/index.ts:2468
Athenaextensions/sonance-cortex/index.ts:2488
CortexDocument generationdocument-generationchosen at runtimetext3not routed
Cortexcore/cortex/hermes/workflows/ralph_setup/workflow.py:1496
Cortexcore/cortex/hermes/workflows/teams_deploy/workflow.py:1068
Cortexcore/cortex/hermes/workflows/workbook_setup/ai_context_generator.py:178
sales-reports-v2Executive summary generationexecutive-summary-generationclaude-sonnet-4-20250514, claude-sonnet-4-6text3not routed
sales-reports-v2src/app/api/reports/[id]/generate-summary/route.ts:135
sales-reportsapp/api/generate-global-summary/route.ts:324
sales-reportsapp/api/generate-summary/route.ts:181
peoplesoft_analysisunlinkedGenerate SQL via Claudenl-to-sql-generationchosen at runtimetext3not routed
Calls Claude via the Anthropic SDK with prompt caching to generate SQL for a user's natural-language question, including continuation handling for truncated output.
peoplesoft_analysisscripts/chatbot_prototype.py:1869
peoplesoft_analysisscripts/chatbot_prototype.py:1879
peoplesoft_analysisscripts/chatbot_prototype.py:1906
CortexImage generationimage-generationchosen at runtimetext, image3not routed
Cortexcore/cortex/hermes/mcps/builtin/google_media/client.py:191
Cortexcore/cortex/hermes/mcps/builtin/google_media/client.py:301
Cortexsupabase/functions/generate-avatar/index.ts:3
logistics-agent-v3Agentic chat streamingagentic-chat-streamingchosen at runtimetext2not routed
logistics-agent-v3web/app/api/demo/chat/route.ts:113
logistics-agent-v3web/lib/fluid/loop.ts:118
sonance-order-automation-2Agentic support chatagentic-support-chatclaude-sonnet-4-20250514, claude-sonnet-4-6text2not routed
sonance-order-automation-2app/api/agent/chat/route.ts:275
sonance-order-automation-2opal-engine/channel-monitor.mjs:66
component-libraryAnalyze screenshot for search termsscreenshot-search-analysisclaude-sonnet-4-20250514text2not routed
Uses Claude vision to extract visible text, likely component names and code patterns from a screenshot to aid source file discovery.
component-librarysrc/app/api/sonance-vision-apply/route.ts:1417
component-librarysrc/app/api/sonance-vision-edit/route.ts:1076
UnitreeAnswer web search queryweb-search-answergpt-4o-minitext2not routed
Uses OpenAI Responses API with web search tool to fetch a concise answer to a user's query for the talk page.
Unitreebob-dashboard/src/app/api/search/route.ts:25
Unitreevoice/realtime.py:84
how-we-win-studio-v2Claim draftingclaim-draftingclaude-3-5-sonnet-20241022text2not routed
how-we-win-studio-v2app/api/ai/draft-claim-stream/route.ts:228
how-we-win-studio-v2lib/ai-client.ts:78
UnitreeConversational agent configconversational-agent-configchosen at runtimeaudio2not routed
Unitreevoice/setup_agent.py:134
Unitreevoice/setup_agent.py:168
peoplesoft_analysisunlinkedDescribe table/view via Ollamaschema-object-descriptionchosen at runtimetext2not routed
Sends a table or view prompt to local Ollama and parses a JSON description (summary, purpose, module, observations).
peoplesoft_analysispipeline/ai_analyzer.py:98
peoplesoft_analysispipeline/ai_table_descriptions.py:936
sonance-order-automation-2Document extractiondocument-extractionclaude-sonnet-4-20250514, claude-sonnet-4-6text2not routed
sonance-order-automation-2app/api/wizard/analyze-samples/route.ts:116
sonance-order-automation-2opal-engine/order-extractor.mjs:134
CortexDocument understandingdocument-understandingchosen at runtimetext2not routed
Cortexcore/cortex/noesis/document_understanding.py:332
Cortexcore/cortex/noesis/meeting_understanding.py:246
ElmounlinkedEmbed text for memory storagetext-embeddingchosen at runtimeembedding2not routed
Generates an OpenAI embedding vector for text so it can be stored/searched in the LanceDB memory table.
Elmoextensions/memory-lancedb/index.ts:174
Elmosrc/memory/embeddings-voyage.ts:34
iportio-agentunlinkedGenerate agent chat replysales-thread-agent-responsechosen at runtimetext2not routed
Calls Claude with order context, conversation history, and tool definitions to generate the AI agent's (Ira) response in a sales Teams thread.
iportio-agentsrc/lib/teams/conversation.ts:224
iportio-agentsrc/lib/teams/conversation.ts:241
component-libraryGenerate vision-based code patchesvision-code-patch-generationclaude-sonnet-4-20250514text2not routed
Sends conversation messages plus a vision system prompt to Claude to generate JSON code modifications/patches based on screenshot analysis, including retry feedback for hallucinated patches.
component-librarysrc/app/api/sonance-vision-apply/route.ts:3846
component-librarysrc/app/api/sonance-vision-edit/route.ts:2875
ElmounlinkedGenerate/edit image with Geminiimage-generationgemini-3-pro-image-previewtext, image2not routed
Calls Google's Gemini image-preview model to generate or edit an image from a prompt and optional input images.
Elmoskills/nano-banana-pro/scripts/generate_image.py:129
Elmoskills/openai-image-gen/scripts/gen.py:87
component-libraryIdentify design problemsdesign-problem-analysisclaude-sonnet-4-20250514text2not routed
Uses Claude vision to analyze a screenshot and identify specific UI/UX issues (hierarchy, spacing, alignment, feedback, grouping) for a given user request.
component-librarysrc/app/api/sonance-vision-apply/route.ts:1509
component-librarysrc/app/api/sonance-vision-edit/route.ts:1168
athena-v2Image generationimage-generationgemini-3-pro-image-previewtext, image2not routed
Athenaskills/nano-banana-pro/scripts/generate_image.py:129
Athenaskills/openai-image-gen/scripts/gen.py:87
BloomLiving stage designliving-stage-designclaude-sonnet-5text2not routed
Bloomlib/anthropic.js:434
Bloomlib/anthropic.js:524
atlasLLM gateway proxyllm-gateway-proxychosen at runtimetext2not routed
atlassrc/lib/gateway.ts:407
atlassrc/lib/gateway.ts:418
atlasMigration doc examplemigration-doc-examplegpt-4otext2not routed
atlassrc/lib/migrationPrompt.ts:306
atlassrc/lib/migrationPrompt.ts:322
atlasModel access probemodel-access-probechosen at runtimetext2not routed
atlassrc/lib/modelProbe.ts:56
atlassrc/lib/modelProbe.ts:65
BloomMotion direction generationmotion-direction-generationclaude-sonnet-5text2not routed
Bloomlib/anthropic.js:30
Bloomlib/anthropic.js:74
atlasProvider endpoint researchprovider-endpoint-researchclaude-opus-5, mistral.aitext2not routed
atlassrc/lib/providerResearch.ts:52
atlassrc/lib/providerResearch.ts:85
CortexRAG answer synthesisrag-answer-synthesischosen at runtimetext2not routed
Cortexcore/cortex/noesis/answer.py:297
Cortexcore/cortex/noesis/resolve.py:205
dispatchRepo app summarizationrepo-app-summarizationchosen at runtimetext2not routed
dispatchapp/api/analyze/route.ts:154
dispatchapp/api/workspaces/[id]/draft/route.ts:24
ElmounlinkedSearch web via xAI Grokweb-searchchosen at runtimetext2not routed
Defines endpoint/config constants for issuing web search queries through xAI's Grok responses API.
Elmosrc/agents/tools/web-search.ts:32
Elmosrc/agents/tools/web-search.ts:458
component-librarySelect best matching filefile-selection-from-screenshotclaude-sonnet-4-20250514text2not routed
Uses Claude vision with a screenshot and candidate file list to pick the file most likely containing the UI element the user wants to modify.
component-librarysrc/app/api/sonance-vision-apply/route.ts:1776
component-librarysrc/app/api/sonance-vision-edit/route.ts:1435
ElmounlinkedStream ElevenLabs TTS audiotts-elevenlabschosen at runtimeaudio2not routed
Opens a streaming HTTP connection to ElevenLabs text-to-speech API and plays back PCM audio on the Android talk mode.
Elmoapps/android/app/src/main/java/ai/openclaw/android/voice/TalkModeManager.kt:947
Elmosrc/tts/tts-core.ts:552
BloomStreamed text generationstreamed-text-generationchosen at runtimetext2not routed
Bloomlib/anthropic.js:613
Bloomlib/anthropic.js:663
Cortex-DashboardStudio chat assistantstudio-chat-assistantclaude-sonnet-4-5-20250929text2not routed
Cortex-Dashboardapp/api/studio/chat/route.ts:94
Cortex-Dashboardapp/api/studio/prd/chat/route.ts:269
athena-v2Synthesize speech via OpenAI TTStts-openaichosen at runtimeaudio2not routed
Calls OpenAI's audio speech endpoint to convert text to PCM audio for voice call playback.
Athenaextensions/voice-call/src/providers/tts-openai.ts:111
Athenasrc/tts/tts-core.ts:612
ElmounlinkedSynthesize speech via OpenAI TTStts-openaichosen at runtimeaudio2not routed
Calls OpenAI's audio/speech endpoint to generate PCM speech audio from text for voice call playback.
Elmoextensions/voice-call/src/providers/tts-openai.ts:111
Elmosrc/tts/tts-core.ts:612
athena-v2Text embeddingtext-embeddingchosen at runtimeembedding2not routed
Athenaextensions/memory-lancedb/index.ts:174
Athenasrc/memory/embeddings-voyage.ts:34
athena-v2Tts elevenlabstts-elevenlabschosen at runtimeaudio2not routed
Athenaapps/android/app/src/main/java/ai/openclaw/android/voice/TalkModeManager.kt:947
Athenasrc/tts/tts-core.ts:552
athena-v2Web search agentweb-search-agentchosen at runtimetext2not routed
Athenasrc/agents/tools/web-search.ts:32
Athenasrc/agents/tools/web-search.ts:458
BloomAnalyze logo for recreationlogo-analysisclaude-sonnet-5text1not routed
Calls Claude vision to extract a logo's exact text, colors, style and layout for faithful recreation.
Bloomlib/anthropic.js:142
peoplesoft_analysisunlinkedAnalyze process doc with images via Claudedocument-multimodal-analysischosen at runtimetext1not routed
Calls Claude with optional embedded images and cached system prompt to analyze process/business documentation, returning text and token usage.
peoplesoft_analysispipeline/pb_analyze.py:110
peoplesoft_analysisunlinkedAnalyze table cluster via Ollamaschema-cluster-analysischosen at runtimetext1not routed
Sends a cluster-of-tables prompt to Ollama and parses JSON describing the cluster's business process and cross-module dependencies.
peoplesoft_analysispipeline/ai_cluster_analysis.py:425
component-libraryApply AI code editai-code-edit-generationclaude-sonnet-4-20250514text1not routed
Sends source code, user request and chat history to Claude to generate a modified file plus explanation and preview CSS for a UI edit request.
component-librarysrc/app/api/sonance-ai-edit/route.ts:381
peoplesoft_analysisunlinkedAudit program description via Ollamaprogram-description-auditchosen at runtimetext1not routed
Runs a secondary Ollama (Llama) pass over the same program prompt as an audit check against the Claude output.
peoplesoft_analysispipeline/ai_describe_programs.py:140
CortexAudit stated role vs activitydiscrepancy-auditchosen at runtimetext1not routed
Calls Claude to compare a person's stated role/data against observed activity and flag discrepancies.
Cortexcore/cortex/noesis/discrepancy_audit.py:247
peoplesoft_analysisunlinkedBatch embed texts via Ollamaembedding-clientchosen at runtimeembedding1not routed
Core reusable client method posting a batch of texts to Ollama's embed endpoint with retry logic, returning vector embeddings.
peoplesoft_analysispipeline/ollama_client.py:51
athena-v2Chat reply via CLI agentchat-completionchosen at runtimetext1not routed
Sends accumulated conversation messages to Anthropic's Claude model through the Athena gateway to generate a conversational reply for the CLI agent.
athena-v2services/anthropic-sdk-agent/bin/run.mjs:51
athena-v2Classify GitHub issue/PRissue-classificationchosen at runtimetext1not routed
Calls OpenAI's responses API with a JSON schema to classify an issue/PR as bug/enhancement, support request, or skill-only.
Athenascripts/label-open-issues.ts:610
ElmounlinkedClassify GitHub issue/PRissue-classificationchosen at runtimetext1not routed
Sends issue/PR text to OpenAI's responses API to classify it as bug/enhancement, support request, or skill-only via structured JSON output.
Elmoscripts/label-open-issues.ts:610
iportio-agentunlinkedClassify or research order/companyorder-classification-researchchosen at runtimetext1not routed
Runs a Claude model with web search tooling to classify a purchaser or research company/purchaser profile info, logging the run and tool calls to the database.
iportio-agentsrc/lib/analysis/engine.ts:58
sonance-order-automation-2Classify PDF document typedocument-classificationclaude-sonnet-4-20250514text1not routed
Sends extracted PDF text to Claude to determine whether the document is a purchase order versus invoice/quote/other with a confidence score.
sonance-order-automation-2opal-engine/order-classifier.mjs:29
peoplesoft_analysisunlinkedClassify table sensitivity via Claudetable-sensitivity-classificationchosen at runtimetext1not routed
Calls Claude with a cached system prompt describing sensitivity groups to classify a batch of PeopleSoft tables for access control.
peoplesoft_analysisscripts/classify_sensitivity.py:137
athena-v2Count tokens for messagestoken-countingchosen at runtimetoken counting1not routed
Provides a persistent Anthropic SDK client configured for the /v1/messages/count_tokens endpoint to estimate token usage before/around inference calls.
athena-v2src/lib/anthropic.ts:76
sonance-crm-liveCRM chat assistant replycrm-chat-assistantchosen at runtimetext1not routed
Sends user chat messages plus live CRM context (tasks, pipeline, reps, dormant accounts) to Claude to generate a conversational assistant response.
sonance-crmserver/routes/chat.js:198
peoplesoft_analysisunlinkedDescribe program via Claudeprogram-descriptionchosen at runtimetext1not routed
Calls Claude (Sonnet) with cached system prompt to summarize a PeopleSoft Application Engine program.
peoplesoft_analysispipeline/ai_describe_programs.py:120
BloomDesign logo motion treatmentslogo-motion-designclaude-sonnet-5text1not routed
Calls Claude vision to art-direct three distinct animated motion treatments for a logo mark.
Bloomlib/anthropic.js:362
BloomDesign mural scene from themescene-prompt-designclaude-sonnet-5text1not routed
Streams a Claude request to design a bioluminescent mural scene and motion prompt from a brand's viz theme.
Bloomlib/anthropic.js:889
phase-gateunlinkedDiagnose and fix failing testautomated-test-fixingchosen at runtimetext1not routed
Sends failing test details and file/command context to Claude to generate a code fix, shell command, or no-fix diagnosis.
phase-gatesrc/fixer.ts:84
Analysis-DashboardDiagnose NL→SQL correction reportcorrection-triagechosen at runtimetext1not routed
Classifies a human-filed correction against generated SQL into a defect category and produces a root-cause diagnosis for maintainers.
Analysis-Dashboardlib/data/feedback-enrich.ts:49
CortexDirect Claude code review callcode-review-generationchosen at runtimetext1not routed
Falls back to a direct HTTP call to Anthropic's messages API to generate an AI code review when the Apollo proxy is unavailable.
Cortexcore/cortex/hermes/workflows/clean_code/ai_reviewer.py:326
CortexDirect Claude finish-line aggregationanalysis-aggregationchosen at runtimetext1not routed
Falls back to a direct Anthropic API call to aggregate finish-line analysis results into a summarized text.
Cortexcore/cortex/hermes/workflows/finish_line_analysis/workflow.py:1618
CortexDirect Claude import rewrite callcode-transformationchosen at runtimetext1not routed
Falls back to a direct HTTP call to Anthropic's messages API to rewrite import statements in code when Apollo is unavailable.
Cortexcore/cortex/hermes/workflows/clean_code/import_rewriter.py:325
BloomDirect logo edit instructionlogo-edit-directionclaude-sonnet-5text1not routed
Calls Claude with the current logo image and conversation to turn a user's request into a concrete edit instruction.
Bloomlib/anthropic.js:247
Analysis-DashboardDraft API registry metadataapi-metadata-generationchosen at runtimetext1not routed
Generates name, description, tags, category, column definitions and changelog metadata for a published data API from its SQL and sample rows.
Analysis-Dashboardlib/data/api-metadata.ts:344
iportio-agentunlinkedDraft intro sales emailsales-email-draftingchosen at runtimetext1not routed
Uses Claude to draft a personalized introductory email to a new purchaser based on order context, then creates an Outlook draft.
iportio-agentsrc/lib/teams/conversation.ts:154
command-center-demoDraft message replydraft-reply-generationclaude-sonnet-4-20250514text1not routed
Streams an AI-generated reply to an email/chat message using sender, relationship, and vault context along with user guidance.
command-center-cortexsrc/app/api/ai/draft-reply/route.ts:194
CortexEdit image with Geminiimage-editingchosen at runtimetext1not routed
Uses Google's Gemini model to edit a source image (with optional mask and references) based on a text prompt.
Cortexcore/cortex/hermes/mcps/builtin/google_media/client.py:226
BloomEnhance raw image promptprompt-enhancementclaude-sonnet-5text1not routed
Calls Claude to expand a user's raw idea into a richer, optionally brand-tuned image generation prompt.
Bloomlib/anthropic.js:1051
Analysis-DashboardEnrich raw idea into structured ticketticket-enrichmentchosen at runtimetext1not routed
Converts a raw feature idea into a structured ticket (title, problem, acceptance criteria, size, labels, priority) grounded in the project's repo knowledge.
Analysis-Dashboardlib/data/idea-agent.ts:187
Analysis-DashboardExplore repo and submit resultrepo-exploration-agentchosen at runtimetext1not routed
Runs a streaming tool-use loop where the agent reads/searches repo files and eventually submits a structured plan or analysis result.
Analysis-Dashboardlib/data/idea-agent.ts:400
iport-certificationsExtract certification data from documentdocument-extractionclaude-sonnet-4-5-20250929text1not routed
Sends a certification document/image plus a structured prompt to Claude to extract fields like cert number, type, standard, and expiry dates into JSON.
iport-certificationssrc/app/api/v1/certifications/analyze/route.ts:215
CortexExtract entities from contententity-extractionchosen at runtimetext1not routed
Calls Claude via the Anthropic SDK to extract structured entities from content for the knowledge graph, under a spend budget.
Cortexcore/cortex/noesis/entity_extraction.py:414
iport-npi-platform-v2Extract install steps from manualdocument-extractionclaude-haiku-4-5-20251001text1not routed
Sends extracted PDF manual text to Claude to parse and structure installation steps into a JSON array of titled instructions with cautions.
iport-npi-platformsrc/app/api/manual/extract-steps/route.ts:50
logistics-agent-v3Extract intake fields from messagestructured-field-extractionchosen at runtimetext1not routed
Calls Claude with a forced tool call to extract structured shipment intake fields (questions, SKUs, lines) from user messages/conversation.
logistics-agent-v3web/lib/agent/llm.ts:203
UnitreeExtract structured facts from contentdocument-extractionclaude-haiku-4-5-20251001text1not routed
Calls Claude to extract a summary, key facts, and FAQs from scraped source content for Bob's knowledge base.
Unitreebob-dashboard/src/lib/extract-with-llm.ts:48
BloomGenerate asset placard storyasset-story-generationchosen at runtimetext1not routed
Calls Claude to write a short gallery-placard style story describing a finished generative asset.
Bloomlib/anthropic.js:931
CortexGenerate changelog entrychangelog-generationchosen at runtimetext1not routed
Calls the internal Anthropic-compatible proxy to generate a structured changelog entry (title, description, risk) from PR/commit text.
Cortexscripts/changelog/publish_changelog_entry.py:118
CortexGenerate codebase documentationdocumentation-generationchosen at runtimetext1not routed
Sends messages, system prompt and tool definitions to Anthropic's messages API to generate a round of codebase documentation output.
Cortexcore/cortex/hermes/workflows/codebase_docs/doc_generator.py:368
Analysis-DashboardGenerate dashboard visual designdashboard-design-generationchosen at runtimetext1not routed
Produces a structured set of dashboard visuals (charts/KPIs) and filters for given datasets via a tool call.
Analysis-Dashboardlib/data/design-agent.ts:177
sonance-order-automation-2Generate extraction prompt from samplesprompt-generationclaude-sonnet-4-6text1not routed
Sends sample PDFs plus a meta-prompt to Claude to auto-generate a customer-specific extraction prompt for order header/line/account routing.
sonance-order-automation-2app/api/wizard/generate-prompt/route.ts:71
command-center-demoGenerate meeting briefingmeeting-prep-briefingclaude-sonnet-4-20250514text1not routed
Uses Claude with web search tools to research attendees/companies and produce a structured JSON meeting prep briefing with summary, insights, talking points, and risks.
command-center-cortexsrc/app/api/ai/meeting-prep/route.ts:169
command-center-demoGenerate morning briefmorning-brief-generationclaude-sonnet-4-20250514text1not routed
Generates a structured JSON daily morning brief summarizing the user's schedule and relationship context.
command-center-cortexsrc/app/api/ai/morning-brief/route.ts:109
UnitreeGenerate persona-based prompt textpersona-text-generationclaude-haiku-4-5-20251001text1not routed
Calls Claude to generate a filled prompt response styled with Bob's persona, tone, and length settings.
Unitreebob-dashboard/src/app/api/prompts/generate/route.ts:128
CortexGenerate plain-language PR summarypr-summarizationclaude-haiku-4-5-20251001text1not routed
Calls Claude Haiku to produce a non-technical summary of a code fix PR for a non-technical stakeholder.
Cortexscripts/demo_feedback_pipeline.py:366
phase-gateunlinkedGenerate test plan from roadmaptest-plan-generationchosen at runtimetext1not routed
Sends a roadmap section to Claude to generate a structured JSON test plan for a project phase.
phase-gatesrc/generator.ts:38
AI-IntranetGenerate vibe-coding project guideguide-generationclaude-3-5-sonnet-20241022text1not routed
Calls Claude to generate a structured step-by-step SOP guide (or revision guidance) with prompts for creating or improving a Claude Artifact project based on user input and admin feedback.
AI-Intranetapp/api/vibe-coding/generate-guide/route.ts:83
BloomInterpret reference image into promptsreference-image-interpretationclaude-sonnet-5text1not routed
Calls Claude vision to read a reference image's feeling and invent three original scene prompts sharing it.
Bloomlib/anthropic.js:1200
phase-gateunlinkedJudge test pass/failtest-result-judgingchosen at runtimetext1not routed
Uses Claude to evaluate whether actual command output matches the expected test result, returning PASS or FAIL.
phase-gatesrc/runner.ts:160
dispatchLabel LLM call sitesllm-callsite-labelingchosen at runtimetext1not routed
Sends code regions and a manifest of detected call sites to Claude to classify/describe each LLM call site with structured JSON output.
dispatchlib/callsite-scan.ts:248
athena-v2List ElevenLabs voicestts-voice-listingchosen at runtimeaudio1not routed
Fetches the list of available ElevenLabs voices for the talk-voice extension's voice picker.
Athenaextensions/talk-voice/index.ts:27
ElmounlinkedList ElevenLabs voicestts-voice-listingchosen at runtimeaudio1not routed
Fetches the available voice catalog from ElevenLabs' voices API for the talk-voice extension.
Elmoextensions/talk-voice/index.ts:27
component-libraryLocate element line numbercode-element-locationclaude-sonnet-4-20250514text1not routed
Asks Claude to find the line number in a source file where a specific UI element is defined, given file content and element metadata.
component-librarysrc/app/api/sonance-vision-apply/route.ts:967
BloomLocate tagline bounding boxlogo-text-measurementclaude-sonnet-5text1not routed
Calls Claude vision to measure the bounding box and style of a logo's tagline text for re-typesetting.
Bloomlib/anthropic.js:315
national-accounts-crmMap import columns to fieldsimport-column-mappingclaude-haiku-4-5-20251001text1not routed
Uses Claude to map spreadsheet column headers to target CRM database fields based on sample data.
national-accounts-crmapp/api/import/analyze/route.ts:21
green-crmMap spreadsheet columns to CRM fieldscsv-column-mappingclaude-haiku-4-5-20251001text1not routed
Uses Claude to infer which spreadsheet column headers correspond to target CRM import fields based on sample data.
green-crmapp/api/import/analyze/route.ts:21
national-accounts-crmNormalize spreadsheet rowsimport-data-normalizationchosen at runtimetext1not routed
Uses Claude to convert messy spreadsheet rows into clean, structured CRM records matching target fields.
national-accounts-crmapp/api/import/smart/route.ts:30
Analysis-DashboardParameterize verified SQL querysql-parameterizationchosen at runtimetext1not routed
Rewrites a QA'd SQL query to add optional NULL-safe named filter parameters for the API registry.
Analysis-Dashboardlib/data/api-metadata.ts:298
command-center-demoParse focus rules from textnl-to-structured-rulesclaude-sonnet-4-20250514text1not routed
Converts a user's natural language description into structured focus/notification rules matching a defined schema.
command-center-cortexsrc/app/api/ai/parse-focus-rules/route.ts:43
CortexProxy Anthropic chat completionllm-chat-completion-proxychosen at runtimetext1not routed
Sends a chat/messages request to Anthropic via the official SDK on behalf of Apollo proxy clients and returns the response as a dict.
Cortexcore/cortex/apollo/proxy/client.py:270
Analysis-DashboardRefine idea in chat turnidea-refinement-chatchosen at runtimetext1not routed
Generates a concise assistant reply in a chat thread helping refine a feature idea into a buildable spec.
Analysis-Dashboardlib/data/idea-agent.ts:228
Analysis-DashboardReformulate data request for NL→SQL enginequery-request-distillationchosen at runtimetext1not routed
Rewrites a vague or verbose user request into a precise, self-contained ask the PeopleSoft NL→SQL engine can answer.
Analysis-Dashboardlib/data/build-agent.ts:222
CortexRerank search candidatesresult-rerankingchosen at runtimererank1not routed
Calls Voyage AI's rerank endpoint to reorder candidate documents by relevance to a search query.
Cortexcore/cortex/noesis/embeddings.py:190
atlasResearch model pricing metadatamodel-metadata-enrichmentclaude-opus-5text1not routed
Calls Anthropic's Messages API with web search to find current pricing and a 'best for' description for a batch of models, returning structured proposals via a report tool.
atlassrc/lib/modelEnrich.ts:118
peoplesoft_analysisunlinkedReview remediation proposalproposal-accuracy-reviewchosen at runtimetext1not routed
Sends a data-quality proposal and related context to Claude to get a strict JSON accuracy verdict.
peoplesoft_analysispipeline/accuracy_reviewer.py:253
command-center-demoRoute captured content into vaultknowledge-vault-routingclaude-sonnet-4-20250514text1not routed
Determines where new captured content (email/message) belongs within a personal knowledge vault, returning a structured routing plan as JSON.
command-center-cortexsrc/app/api/ai/vault-capture/route.ts:53
Analysis-DashboardRun agent tool-use loop turndata-agent-tool-loopchosen at runtimetext1not routed
Streams a chat turn with tool access (SOQL/SQL execution tools) letting the agent iteratively query data sources and submit a final result.
Analysis-Dashboardlib/data/build-agent.ts:558
Cortex-DashboardRun agentic vibe-coding chat turnstudio-agentic-tool-chatclaude-sonnet-4-5-20250929text1not routed
Calls Anthropic Claude with tool definitions in an agentic loop to generate assistant responses and tool calls for the vibe-coding studio assistant.
Cortex-Dashboardapp/api/studio/vibe/chat/route.ts:377
athena-v2Stream chat completion from Ollamachat-completion-streamingchosen at runtimetext1not routed
Sends chat messages and tool definitions to a local/remote Ollama server and streams back model responses.
Athenasrc/agents/ollama-stream.ts:287
ElmounlinkedStream chat completion from Ollamachat-completionchosen at runtimetext1not routed
Sends a chat/tool-calling request to a local Ollama server's /api/chat endpoint and streams the model's response.
Elmosrc/agents/ollama-stream.ts:287
peoplesoft_analysisunlinkedStream diagnosis narrationresult-narrationchosen at runtimetext1not routed
Streams a natural-language diagnosis from local Ollama explaining why a query returned zero rows, based on probe results.
peoplesoft_analysispipeline/narrate.py:534
CortexStream knowledge-builder chat responseagentic-chat-completionchosen at runtimetext1not routed
Streams a Claude chat completion with tool use and extended thinking to power the interactive knowledge-builder chat assistant.
Cortexcore/cortex/praxis/api/routes/chat.py:2603
ElmounlinkedStream real-time speech transcriptionrealtime-speech-to-textchosen at runtimeaudio1not routed
Opens an OpenAI realtime WebSocket session configured for live speech-to-text transcription during voice calls.
Elmoextensions/voice-call/src/providers/stt-openai-realtime.ts:113
athena-v2Stream realtime speech transcriptionrealtime-speech-to-textchosen at runtimeaudio1not routed
Opens a WebSocket to OpenAI's realtime transcription API and configures the session for live speech-to-text during voice calls.
Athenaextensions/voice-call/src/providers/stt-openai-realtime.ts:113
Analysis-DashboardStructure context-pack query asksquery-spec-structuringchosen at runtimetext1not routed
Turns each parsed spec query into a clean, DGX-ready ask, dropping app-only fields and raising clarification questions.
Analysis-Dashboardlib/data/build-agent.ts:340
BloomSubmit ComfyUI video workflowcomfyui-video-generationchosen at runtimevideo1not routed
Submits an injected text/seed-parameterized workflow to the local ComfyUI server to generate a video.
Bloomlib/comfy.js:70
BloomSubmit fal.ai generation jobfal-job-submissionchosen at runtimeimage1not routed
Generic submission of any fal.ai endpoint job (image or video) to the fal queue API.
Bloomlib/providers/fal.js:18
BloomSubmit Replicate prediction jobreplicate-job-submissionchosen at runtimeimage1not routed
Submits an image generation job to the Replicate predictions API.
Bloomlib/providers/replicate.js:21
smart-bookmark-agentSummarize bookmarked pagebookmark-summarizationclaude-sonnet-4-5text1not routed
Generates a title, summary, and tags for a saved bookmark based on its URL and page text.
smart-bookmark-agentlib/summarize.ts:19
sonance-formsSummarize survey answer themessurvey-theme-summarizationclaude-sonnet-5text1not routed
Sends a batch of free-text answers for a specific form question to Claude to extract common themes for form response insights.
sonance-formsapp/api/forms/[id]/insights/route.ts:148
hr-survey-trackerSummarize survey responsessurvey-response-summarizationchosen at runtimetext1not routed
Generates an AI-written narrative summary of aggregated HR survey responses filtered by type, audience, and date range for reporting.
hr-survey-trackerapp/api/summaries/generate/route.ts:164
marketing-feedbackSummarize survey responsessurvey-response-summarizationclaude-haiku-4-5-20251001text1not routed
Sends collected survey respondent answers to Anthropic to generate a structured executive summary with sentiment, themes, quotes, and action items.
marketing-feedbacksrc/app/api/responses/ai-summary/route.ts:81
athena-v2Synthesize speech via Edge TTStts-edgechosen at runtimeaudio1not routed
Uses Microsoft Edge's TTS engine to synthesize speech from text and save it to an output file.
Athenasrc/tts/tts-core.ts:661
ElmounlinkedSynthesize speech via Edge TTStts-edgechosen at runtimeaudio1not routed
Uses Microsoft Edge's TTS service to convert text to speech audio and save it to a file.
Elmosrc/tts/tts-core.ts:661
sonance-order-automation-2Transcribe audio recordingaudio-transcriptionchosen at runtimeaudio1not routed
Sends an audio blob to OpenAI's Whisper API to produce a text transcript for the Prompt Builder feature.
sonance-order-automation-2lib/ai/openai.ts:31
athena-v2Transcribe audio via Deepgramaudio-transcriptionchosen at runtimeaudio1not routed
Sends raw audio bytes to Deepgram's listen endpoint and returns the transcribed text.
Athenasrc/media-understanding/providers/deepgram/audio.ts:30
ElmounlinkedTranscribe audio via Deepgramspeech-to-textchosen at runtimeaudio1not routed
Posts raw audio bytes to Deepgram's listen endpoint and extracts the transcribed text.
Elmosrc/media-understanding/providers/deepgram/audio.ts:30
national-accounts-crmTranslate NL question to SQLnl-to-sql-generationchosen at runtimetext1not routed
Uses Claude to convert a plain-English question into a single read-only SQL SELECT query against the CRM schema.
national-accounts-crmapp/api/data/query/route.ts:42
CortexTriage user feedbackticket-triageclaude-haiku-4-5-20251001text1not routed
Calls Claude Haiku to classify and triage an incoming user feedback item into type, severity, and effort.
Cortexscripts/demo_feedback_pipeline.py:98
peoplesoft_analysisunlinkedValidate table relationship via Ollamarelationship-validationchosen at runtimetext1not routed
Sends a relationship-validation prompt to Ollama and parses a JSON verdict, retrying on timeout or malformed JSON.
peoplesoft_analysispipeline/ai_relationship_validation.py:134
Dispatch generated 8/4/2026, 12:02:46 AM · cached in Atlas 8/4/2026, 12:02:46 AM