Integration

orq.ai integration

orq.ai is an LLMOps platform and AI gateway. With this API an agent can call 500+ models through the gateway (chat completions, responses, legacy completions, embeddings, rerank, moderation, image generation, OCR), invoke configured deployments, manage agents and their cron schedules, prompts, tools, datasets and datapoints, knowledge bases (datasources, chunks, search), memory stores, evaluators (including running one), files, identities and projects, redact or detect PII, chunk text, attach feedback and span annotations, and read logs, traces, trace threads and the model catalog. Model requests draw on the workspace's orq.ai credits at each model's token pricing (or on a connected provider key). Not exposed: streaming routes, deprecated routes (legacy agent responses and tasks, human review sets, classify, webhook query), API keys, management keys, webhooks (their signing secret), budgets, routing and guardrail rules, smart routers, policies, workspaces, workspace security, settings and enabled models, users and audit logs, custom model registration (provider credentials), MCP servers and gateways, external agents, skills, the hub, file systems, file content download, alerts, monitors, notifiers, annotation queues, saved views, telemetry, reporting, log and trace aggregation, facets and patterns, web search, routing decisions, audio, image edits and variations. Schedule variables and HTTP evaluator headers and payloads are not returned. Authentication is an orq.ai API key sent as a bearer token in the Authorization header; a project-scoped key only reaches its own project.

Connect orq.ai How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in orq.ai.Built from the 141 orq.ai tools below, on the connection you add once.

What you can do with orq.ai

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Chunk, 7 operations

count_chunksCount chunksPOSTRead-only: counts the chunks of a datasource, optionally filtered by text, enabled state or status. With q, chunks not yet embedded may be missing.
create_chunksCreate chunksPOSTAdds 1 to 100 chunks to a datasource. Chunks sent with an embedding are indexed at once; the others are queued for embedding.
delete_chunkDelete chunkDELETEDeletes a chunk from the datasource and its vector index.
delete_chunksDelete chunksDELETEDeletes up to 100 chunks by id and reports ids that were not found or not deleted.
get_chunkGet chunkGETRetrieves a chunk by its chunk identifier.
list_chunksList chunksGETReturns chunks using cursor pagination, with optional text and processing-status filters.
search_chunksSearch chunksPOSTRead-only: returns a page of chunks of a datasource, filtered by text, enabled state or processing status (page and limit paging).

Evaluator, 6 operations

create_evaluatorCreate evaluatorPOSTCreates an evaluator under a key in a project. An llm_eval needs prompt, model and mode single, or mode jury with a jury of at least two judges; a python_eval needs code and an output_type of boolean or number.
delete_evaluatorDelete evaluatorDELETEPermanently deletes an evaluator.
duplicate_evaluatorDuplicate evaluatorPOSTCopies an evaluator under a new key and returns the new evaluator id.
get_evaluatorGet evaluatorGETRetrieves an evaluator by id with its type-specific configuration (header values and request payloads of HTTP evaluators are not returned).
list_evaluatorsList evaluatorsGETLists the workspace's evaluators (LLM, Python, HTTP and others) with their type and configuration. HTTP evaluator header values and request payloads are not returned.
run_evaluatorRun evaluatorPOSTRuns an existing evaluator on the data you pass (a conversation, or query, output and reference) and returns its result. LLM evaluators spend credits.

Datasource, 5 operations

create_datasourceCreate datasourcePOSTAdds a datasource to a knowledge base: an empty one for manual chunks, or one built from an uploaded file_id that is then chunked and embedded.
delete_datasourceDelete datasourceDELETEPermanently deletes a datasource and its chunks from a knowledge base.
get_datasourceGet datasourceGETRetrieves a datasource and its current processing status and chunk count.
get_datasource_processing_statusGet datasource processing statusGETReturns a datasource's processing status with queued, completed and failed chunk counts. Poll it after create_datasource until processing ends.
list_datasourcesList datasourcesGETReturns the datasources in a knowledge base. Use cursors to page through results and optional query or status filters to narrow the list.

Document, 5 operations

create_memory_documentCreate memory documentPOSTStores a text document (a remembered fact) in a memory, with optional metadata for filtering.
delete_memory_documentDelete memory documentDELETEPermanently deletes one document from a memory.
get_memory_documentGet memory documentGETRetrieves details of a specific memory document by its ID.
list_memory_documentsList memory documentsGETRetrieves a paginated list of documents associated with a specific memory.
run_ocrRun ocrPOSTExtracts text from a document or image URL with an OCR model, keeping its structure, and returns pages. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).

Schedule, 5 operations

create_agent_scheduleCreate agent schedulePOSTCreates a cron schedule that runs the agent with a fixed payload. Each run spends credits. Only a 6-field cron (seconds and minutes 0), at most hourly, is accepted. Variables are stored as plain text and never returned by the readers.
delete_agent_scheduleDelete agent scheduleDELETEPermanently deletes an agent schedule; it will not run again.
get_agent_scheduleGet agent scheduleGETRetrieves one agent schedule with its cron expression, input payload, active flag and last trigger time. Schedule variables are not returned (they can hold secret values).
list_agent_schedulesList agent schedulesGETLists an agent's cron schedules with cadence, input payload, active flag and trigger counts. Schedule variables are not returned (they can hold secret values).
run_agent_schedule_nowRun agent schedule nowPOSTRuns a schedule's payload against its agent right away (about 10 seconds after the call), outside its cron cadence, which is unchanged. The run spends credits like any agent run.

Version, 5 operations

get_prompt_versionGet prompt versionGETRetrieves a specific version of a prompt by its ID and version ID.
get_tool_versionGet tool versionGETRetrieves one saved version of a tool by version id, as returned by list_tool_versions.
list_deployment_versionsList deployment versionsGETLists saved versions of a deployment's variants, drafts included, newest first. Pass the deployment key or id from list_deployments; page with a version _id in starting_after or ending_before.
list_prompt_versionsList prompt versionsGETReturns a list of your prompt versions. The prompt versions are returned sorted by creation date, with the most recent prompt versions appearing first
list_tool_versionsList tool versionsGETLists the version history of a tool, newest first.

Agent, 4 operations

create_agentCreate agentPOSTCreates an agent with a model, role, instructions and settings (tools, knowledge bases, memory stores). Run it afterwards with create_response and model agent/<key>.
delete_agentDelete agentDELETEPermanently deletes an agent and its configuration from the workspace.
get_agentGet agentGETRetrieve the complete agent manifest by key, including model assignments, tools, knowledge bases, memory stores, and execution parameters.
list_agentsList agentsGETList all agents in the workspace with full configuration details. Supports pagination and sorts agents newest first.

Datapoint, 4 operations

create_datapointsCreate datapointsPOSTAdds datapoints (inputs, messages, expected output) to a dataset, 1 to 5000 per call.
delete_datapointDelete datapointDELETEPermanently deletes a specific datapoint from a dataset.
get_datapointGet datapointGETRetrieves one datapoint of a dataset with its inputs, messages and expected output.
list_datapointsList datapointsGETRetrieves a paginated list of datapoints from a specific dataset.

Dataset, 4 operations

create_datasetCreate datasetPOSTCreates a new dataset in the specified project.
delete_datasetDelete datasetDELETEPermanently deletes a dataset and all its datapoints. This action is irreversible.
get_datasetGet datasetGETRetrieves a specific dataset by its unique identifier
list_datasetsList datasetsGETRetrieves a paginated list of datasets for the current workspace. Results can be paginated using cursor-based pagination.

File, 4 operations

delete_fileDelete fileDELETEPermanently deletes a file and its stored content.
get_fileGet fileGETRetrieves a file's metadata (name, purpose, size, project) by file_id.
list_filesList filesGETReturns a list of the files that your account has access to. orq.ai sorts and returns the files by their creation dates, placing the most recently created files at the top.
rename_fileRename filePATCHRenames a file (its display name).

Identity, 4 operations

create_identityCreate identityPOSTCreates an identity (an end user or customer tracked for usage and budgets) with a unique external_id; if one already exists with that external_id it is returned.
delete_identityDelete identityDELETEPermanently deletes an identity and its identity-scoped budget configuration.
get_identityGet identityGETRetrieves detailed information about a specific identity using their identity ID or external ID from your system.
list_identitiesList identitiesGETRetrieves a paginated list of identities in your workspace. Use pagination parameters to navigate through large identity lists efficiently.

Knowledge, 4 operations

create_knowledge_baseCreate knowledge basePOSTCreates an internal or external knowledge base. Internal knowledge bases embed and index uploaded content; external knowledge bases query the configured external retrieval API.
delete_knowledge_baseDelete knowledge baseDELETEPermanently deletes a knowledge base with all its datasources and chunks.
get_knowledge_baseGet knowledge baseGETRetrieves a knowledge base with its embedding model and retrieval settings.
list_knowledge_basesList knowledge basesGETReturns a list of your knowledge bases. The knowledge bases are returned sorted by creation date, with the most recent knowledge bases appearing first

Log, 4 operations

get_logGet logGETRetrieve a log record by its ULID. Returns 404 if the record does not exist or belongs to another workspace.
get_log_contextGet log contextPOSTRead-only: returns the matching log records just before and after an anchor log id, in chronological order.
list_trace_logsList trace logsGETReturn all log records correlated with a given trace_id. Results are scoped to the authenticated workspace.
query_logsQuery logsPOSTRead-only: runs an OQL log query (fetch logs | filter ... | limit N) over a from/to time range; newest first.

Memory, 4 operations

create_memoryCreate memoryPOSTCreates a memory (an entity such as a user or session) in a memory store, keyed by entity_id.
delete_memoryDelete memoryDELETEPermanently deletes a memory and its documents.
get_memoryGet memoryGETRetrieves details of a specific memory by its ID
list_memoriesList memoriesGETRetrieves a paginated list of memories for the memory store

Memory store, 4 operations

create_memory_storeCreate memory storePOSTCreates a memory store with an embedding configuration, description, optional TTL and project path.
delete_memory_storeDelete memory storeDELETEPermanently delete a memory store, including memories and documents.
get_memory_storeGet memory storeGETRetrieves detailed information about a specific memory store, including its configuration and metadata.
list_memory_storesList memory storesGETRetrieves a paginated list of memory stores in the workspace. Use cursor-based pagination parameters to navigate through the results.

Project, 4 operations

create_projectCreate projectPOSTCreates a project in the workspace. Needs a workspace-scoped management key or an API key with Projects write access; project-scoped API keys cannot.
delete_projectDelete projectDELETEPermanently deletes a project. Needs a workspace-scoped management key or an API key with Projects write access.
get_projectGet projectGETRetrieves the details of an existing project by its unique project ID.
list_projectsList projectsGETLists the workspace's projects, newest first. Page with a project_id in starting_after or ending_before.

Prompt, 4 operations

create_promptCreate promptPOSTCreates a prompt in a project path with a model and messages. A literal {{name}} placeholder must be written {{'{{name}}'}} in a workflow, or the workflow engine blanks it.
delete_promptDelete promptDELETEPermanently deletes a prompt and all its versions.
get_promptGet promptGETRetrieves a prompt by id with its prompt configuration (model, parameters, messages) and metadata.
list_promptsList promptsGETReturns a list of your prompts. The prompts are returned sorted by creation date, with the most recent prompts appearing first

Tool, 4 operations

create_toolCreate toolPOSTCreates a tool (function, JSON schema, HTTP or code) that agents and prompts can call. Header values on an HTTP tool are not returned by the tool readers.
delete_toolDelete toolDELETEPermanently deletes a tool from the workspace.
get_toolGet toolGETRetrieves a tool by id with its type-specific definition. HTTP tool header values are not returned.
list_toolsList toolsGETList all workspace tools. By default returns all tools in one response. Set `limit` for cursor-based pagination.

Deployment, 3 operations

get_deployment_configGet deployment configPOSTRead-only: returns the resolved prompt configuration (model, parameters, messages, tools) a deployment would use for the given key and context, without calling a model.
invoke_deploymentInvoke deploymentPOSTRuns a deployment (a configured prompt and model) by its key and returns the model's reply in choices, with retrievals and usage. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).
list_deploymentsList deploymentsGETLists the workspace's deployments, newest first, with their key and prompt configuration. Page with starting_after and ending_before.

Pii, 3 operations

detect_piiDetect piiPOSTRead-only: reports whether a text contains personally identifiable information, optionally with a per-type entity breakdown.
redact_piiRedact piiPOSTReplaces detected PII in a text with placeholders and returns the redacted text plus the mapping that restore_pii needs.
restore_piiRestore piiPOSTPuts the original values back into a text redacted by redact_pii, using the mapping it returned.

Annotation, 2 operations

annotate_spanAnnotate spanPOSTAttaches annotations (human review values, or a correction of an evaluator output) to a span of a trace.
delete_span_annotationsDelete span annotationsDELETERemoves annotations from a span by their review keys, or removes corrections by the evaluator ids they correct.

Completion, 2 operations

create_chat_completionCreate chat completionPOSTSends a chat conversation to a model through the orq.ai gateway (OpenAI-compatible) and returns the reply in choices, with fallbacks, retries and cache. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).
create_completionCreate completionPOSTLegacy text completion: sends a raw prompt to a completion model through the gateway and returns choices. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).

Facet, 2 operations

list_trace_facetsList trace facetsGETList trace fields that support facet value discovery.
list_trace_facet_valuesList trace facet valuesGETList values and counts for one trace facet field.

Field, 2 operations

list_log_fieldsList log fieldsGETReturn all queryable fields: static columns and dynamic attribute families (attribute.*, resource.*).
list_trace_fieldsList trace fieldsGETList currently supported static trace fields.

Model, 2 operations

list_router_modelsList router modelsGETLists all models available through the AI Router. Returns each model in OpenAI-compatible shape with its provider, ID, and creation timestamp.
list_workspace_modelsList workspace modelsGETLists the models available to the workspace, system and custom, with their enabled status, type and parameters.

Model catalog, 2 operations

get_model_catalog_entryGet model catalog entryGETRetrieves one model catalog entry by its <provider>/<model> id, deprecated models included (check deprecated on the response).
list_model_catalogList model catalogGETLists the models orq.ai offers (deprecated ones excluded), with filters by provider, modality, region, feature and tier. Leave limit out to get the whole catalog; page with starting_after.

Response, 2 operations

create_responseCreate responsePOSTCreates a model response (OpenAI Responses shape). Use model provider/model, or agent/<key> to run an agent; read the result in output. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).
get_responseGet responseGETRetrieves a stored response by id, as returned by create_response (use it to read a background response once its status is completed).

Session, 2 operations

delete_threadDelete threadDELETEPermanently deletes a trace thread.
get_threadGet threadGETRetrieves a trace thread by id with its duration, trace count, usage and tags.

Span, 2 operations

get_trace_spanGet trace spanGETRetrieves one span of a trace with its attributes, as listed by list_trace_spans.
list_trace_spansList trace spansGETList canonical span summaries for a trace.

Thread, 2 operations

count_threadsCount threadsPOSTRead-only: counts trace threads matching the filters.
list_threadsList threadsPOSTRead-only: lists trace threads (conversations grouping traces) matching the filters, page by page.

Trace, 2 operations

get_traceGet traceGETRetrieve one trace summary by trace id.
query_tracesQuery tracesPOSTRead-only: runs an OQL query selecting traces over a from/to time range, paged with page_token.

Capability, 1 operation

get_pii_capabilitiesGet pii capabilitiesGETReports the entity catalog, supported regions, and default thresholds the PII detection service currently exposes, including per-entity default thresholds.

Clear, 1 operation

clear_dataset_datapointsClear dataset datapointsDELETEDelete all datapoints from a dataset. This action is irreversible.

Conversation, 1 operation

get_trace_conversationGet trace conversationGETReturn ordered OpenResponses items from the selected model-call span. Prefers spans with output outside evaluator subtrees unless `span_id` is given.

Embedding, 1 operation

create_embeddingCreate embeddingPOSTReturns embedding vectors for the input text from the chosen embedding model. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).

Feedback, 1 operation

remove_feedbackRemove feedbackPOSTRemoves a feedback entry from a trace by its id.

Image, 1 operation

generate_imageGenerate imagePOSTGenerates images from a text prompt with an image model; returns a temporary provider url or base64 per image. n defaults to 1 (the cheapest). Spends orq.ai credits at the model's token pricing (or uses a connected provider key).

Moderation, 1 operation

create_moderationCreate moderationPOSTClassifies text against content policy categories with a moderation model and returns flagged categories and scores. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).

Offering, 1 operation

list_model_offeringsList model offeringsGETReturns every provider offering of one base model, identified by `<developer>/<stem>` (for example `anthropic/claude-opus-4-7`). Deprecated models are never listed.

Rerank, 1 operation

create_rerankCreate rerankPOSTReranks a list of documents by relevance to a query with a rerank model and returns indices and scores. Spends orq.ai credits at the model's token pricing (or uses a connected provider key).

Tag, 1 operation

list_thread_tagsList thread tagsGETRetrieves all unique thread tags in the workspace.

Text, 1 operation

chunk_textChunk textPOSTSplits text into chunks for RAG with a token, sentence, recursive, semantic, agentic or fast strategy. Semantic and agentic strategies call a model and spend credits.

Showing the first 120 of 141 endpoints. The rest are in the product, searchable from the workflow builder and from chat.

Use orq.ai from Claude Code, Cursor or Codex

Connect orq.ai in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its orq.ai operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect orq.ai

Create your workspace
Sign up free. Nothing to install.
Connect orq.ai once
Add your orq.ai connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.