Integration

Griptape Cloud integration

Griptape Cloud hosts AI apps built with the Griptape framework. Run hosted Structures (agents, pipelines, workflows) and Tools and read their runs, events and logs; chat with no-code Assistants; query or search Knowledge Bases and Retrievers built from Data Sources and refresh those sources; manage Rulesets and Rules that steer LLM behaviour, Threads and Messages that hold conversation memory, and Data Lake buckets and assets. Running Structures, Assistants and Tools spends your Griptape Cloud usage and the model provider usage of the LLM calls they make. Assistant runs, threads and messages hold end-user conversations. Not exposed: API keys, secrets, connections and integrations (they hold third-party tokens), creating or editing Knowledge Bases, Data Sources and Retrievers (their configuration carries database passwords and access tokens), deployments, Functions, model configurations, billing, budgets, usage, licenses, organizations, users, sessions and the streaming event routes. Environment variable values and stored access tokens are stripped from every answer. Authentication is a Griptape Cloud API key sent as a Bearer token.

Connect Griptape Cloud How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Griptape Cloud.Built from the 84 Griptape Cloud tools below, on the connection you add once.

What you can do with Griptape Cloud

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.

Assistant, 5 operations

create_assistantCreate assistantPOSTCreates an Assistant, a ready-made chat agent, from a name plus optional description and the ids of the Knowledge Bases, Retrievers, Rulesets, Tools and Structures it may use. Returns the new Assistant.
delete_assistantDelete assistantDELETEDeletes the Assistant with this id from the organization; the answer has no body.
get_assistantGet assistantGETReturns one Assistant with its description and the ids of the Knowledge Bases, Retrievers, Rulesets, Tools and Structures it uses.
list_assistantsList assistantsGETLists the Assistants (ready-made chat agents configured without code) in the organization with the Knowledge Bases, Retrievers, Rulesets, Tools and Structures each may use, a page at a time.
update_assistantUpdate assistantPATCHChanges an Assistant's name, description or the ids of its Knowledge Bases, Rulesets, Tools or Structures. The docs do not say whether omitted fields keep their values: send every field and complete id list you need.

Bucket, 5 operations

create_bucketCreate bucketPOSTCreates a Data Lake Bucket with a name and returns it with its bucket_id.
delete_bucketDelete bucketDELETEDeletes the Data Lake Bucket with this id; the answer has no body.
get_bucketGet bucketGETReturns one Data Lake Bucket with its name and timestamps.
list_bucketsList bucketsGETLists the Data Lake Buckets in the organization (cloud storage containers for files used by Data Sources and Structures), a page at a time.
update_bucketUpdate bucketPATCHRenames a Data Lake Bucket (name is its only editable field) and returns it.

Message, 5 operations

create_messageCreate messagePOSTAppends a Message to a Thread: the user's input and the LLM's output, with optional metadata. Returns the new Message with its index in the Thread.
delete_messageDelete messageDELETEDeletes one Message from its Thread, so the LLM no longer recalls it; the answer has no body.
get_messageGet messageGETReturns one Message of a Thread: its input, output, metadata and index. Holds end-user conversation content.
list_messagesList messagesGETLists the Messages of a Thread in order (each the user's input and the LLM's output), a page at a time. Holds end-user conversation content.
update_messageUpdate messagePATCHChanges a Message's input, output or metadata, to control what the LLM recalls of the conversation. The docs do not say whether omitted fields keep their values: send every field you need to keep.

Rule, 5 operations

create_ruleCreate rulePOSTCreates a Rule, one instruction for an LLM such as 'Always answer in French', with a name and optional metadata. Add it to a Ruleset with update_ruleset. Returns the new Rule.
delete_ruleDelete ruleDELETEDeletes the Rule with this id from the organization; the answer has no body.
get_ruleGet ruleGETReturns one Rule with its name, instruction text and metadata.
list_rulesList rulesGETLists the Rules in the organization (single instructions given to an LLM), a page at a time; pass ruleset_id to list only the Rules of one Ruleset.
update_ruleUpdate rulePATCHChanges a Rule's name, instruction text or metadata. The docs do not say whether omitted fields keep their values: send every field you need to keep.

Ruleset, 5 operations

create_rulesetCreate rulesetPOSTCreates a Ruleset from a name, an optional alias (your own unique identifier), a description, metadata and the ids of the Rules it contains. Returns the new Ruleset.
delete_rulesetDelete rulesetDELETEDeletes the Ruleset with this id from the organization; the answer has no body.
get_rulesetGet rulesetGETReturns one Ruleset with its alias, description, metadata and the ids of its Rules.
list_rulesetsList rulesetsGETLists the Rulesets in the organization (named sets of Rules that steer LLM behaviour), a page at a time; filter by alias to find one by your own identifier.
update_rulesetUpdate rulesetPATCHChanges a Ruleset's name, alias, description, metadata or Rule ids. The docs do not say whether omitted fields keep their values: send every field you need to keep, and the complete rule_ids list.

Run, 5 operations

create_assistant_runCreate assistant runPOSTSends a message (args) to an Assistant and starts a run, which spends your Griptape Cloud usage and LLM usage. Continue a conversation with thread_id or start one with new_thread. Returns the run; poll get_assistant_run until status is SUCCEEDED.
create_structure_runCreate structure runPOSTStarts a run of a deployed Structure with arguments and optional run-only environment variables. Each run spends your Griptape Cloud usage and the model provider usage of its LLM calls. Returns the run; get_structure_run reads its status and output.
list_assistant_runsList assistant runsGETLists the runs of an Assistant (each message sent to it) with status, input, output and thread, a page at a time, optionally only runs in the given statuses. Holds end-user conversation content.
list_structure_runsList structure runsGETLists the runs of a Structure with their status, arguments, output and timestamps, a page at a time, optionally only runs in the given statuses.
list_tool_runsList tool runsGETLists the runs of a hosted Tool (each activity call) with status, input, output and timestamps, a page at a time, optionally only runs in the given statuses.

Structure, 5 operations

create_structureCreate structurePOSTCreates a Structure from code in a GitHub repository or a ZIP file in a Data Lake bucket; the code needs a Structure Config YAML file. Griptape Cloud then starts deploying it. Returns the new Structure.
delete_structureDelete structureDELETEDeletes the Structure with this id from the organization; the answer has no body.
get_structureGet structureGETReturns one Structure: name, description, code source, Structure Config file, webhook flag, environment variable names and sources (values are not returned) and its latest deployment id.
list_structuresList structuresGETLists the Structures in the organization (Griptape Agents, Pipelines, Workflows or other Python code hosted on Griptape Cloud), a page at a time, with their code source and latest deployment id.
update_structureUpdate structurePATCHChanges a Structure's name, description, code source, config file, webhook flag or environment variables. The docs do not say whether omitted fields keep their values: send every field you need to keep.

Thread, 5 operations

create_threadCreate threadPOSTCreates a Thread to store a conversation history, with a name, an optional alias (your own unique identifier), metadata and optional first messages. Returns the new Thread.
delete_threadDelete threadDELETEDeletes the Thread with this id and its stored conversation; the answer has no body.
get_threadGet threadGETReturns one Thread with its alias, metadata, message count and total message length.
list_threadsList threadsGETLists the Threads (stored conversation histories) in the organization with message counts, a page at a time; filter by alias or by the user who created them. Holds end-user conversation content.
update_threadUpdate threadPATCHChanges a Thread's name, alias, metadata or messages. The docs do not say whether omitted fields keep their values: send every field you need to keep.

Tool, 5 operations

create_toolCreate toolPOSTCreates a hosted Tool from code in a GitHub repository or a ZIP file in a Data Lake bucket; the code needs a Tool Config YAML file naming the tool class and init function. Returns the new Tool.
delete_toolDelete toolDELETEDeletes the hosted Tool with this id from the organization; the answer has no body.
get_toolGet toolGETReturns one hosted Tool: name, description, code source, Tool Config file, environment variable names and sources (values are not returned) and its latest deployment id.
list_toolsList toolsGETLists the hosted Tools in the organization (Griptape framework tools whose activities agents can call over HTTP), a page at a time, with their code source and latest deployment id.
update_toolUpdate toolPATCHChanges a hosted Tool's name, description, code source, config file or environment variables. The docs do not say whether omitted fields keep their values: send every field you need to keep.

Cancel, 4 operations

cancel_assistant_runCancel assistant runPOSTCancels an Assistant run and returns the run with its updated status.
cancel_data_jobCancel data jobPOSTCancels a Data Source refresh job and returns it with its updated status.
cancel_knowledge_base_jobCancel knowledge base jobPOSTCancels a Knowledge Base ingestion job and returns it with its updated status.
cancel_structure_runCancel structure runPOSTCancels a Structure run and returns the run with its updated status.

Asset, 3 operations

delete_assetDelete assetDELETEDeletes one asset from a Data Lake Bucket; the answer has no body.
get_assetGet assetGETReturns one asset of a Data Lake Bucket with its size and timestamps, and its contents when include_contents is true.
list_assetsList assetsGETLists the assets (files and folders) of a Data Lake Bucket with their size, a page at a time; filter by name prefix (an asset path such as docs/) or suffix.

Data connector, 3 operations

delete_data_connectorDelete data connectorDELETEDeletes the Data Source with this id from the organization; the answer has no body.
get_data_connectorGet data connectorGETReturns one Data Source: type, configuration (URLs, file ids, S3 URIs, Confluence domain and account email, without access tokens), transforms and refresh schedule.
list_data_connectorsList data connectorsGETLists the Data Sources in the organization (Web Page, Amazon S3, Google Drive, Confluence, Data Lake or Structure) with their configuration and refresh schedule. Access tokens are never returned.

Data job, 3 operations

create_data_jobCreate data jobPOSTRefreshes a Data Source now, like its Refresh button: starts a data job that ingests its data again and returns the job with its status. Follow it with get_data_job.
get_data_jobGet data jobGETReturns one Data Source refresh job: status (QUEUED, RUNNING, SUCCEEDED, FAILED or CANCELLED), bytes ingested, errors and timestamps.
list_data_jobsList data jobsGETLists the refresh jobs of a Data Source with status, bytes ingested, errors and timestamps, a page at a time, optionally only jobs in the given statuses.

Knowledge base, 3 operations

delete_knowledge_baseDelete knowledge baseDELETEDeletes the Knowledge Base with this id from the organization; the answer has no body.
get_knowledge_baseGet knowledge baseGETReturns one Knowledge Base: type, configuration (without any database connection string), asset paths, transforms, embedding model and refresh schedule.
list_knowledge_basesList knowledge basesGETLists the Knowledge Bases in the organization (vector or hybrid indexes built from Data Sources) with their type, asset paths, embedding model and refresh schedule, a page at a time.

Knowledge base job, 3 operations

create_knowledge_base_jobCreate knowledge base jobPOSTStarts an ingestion job that loads the Knowledge Base's Data Source data again; returns the job with its status. Follow it with get_knowledge_base_job.
get_knowledge_base_jobGet knowledge base jobGETReturns one Knowledge Base ingestion job: status (QUEUED, RUNNING, SUCCEEDED, FAILED or CANCELLED), bytes ingested, errors and timestamps.
list_knowledge_base_jobsList knowledge base jobsGETLists the ingestion jobs of a Knowledge Base with status, bytes ingested, errors and timestamps, a page at a time, optionally only jobs in the given statuses.

Query, 3 operations

list_knowledge_base_queriesList knowledge base queriesGETLists the past queries made against a Knowledge Base with the entries each returned, a page at a time.
query_knowledge_baseQuery knowledge basePOSTQueries a Knowledge Base and returns the individual matching entries (id, score, namespace, metadata) rather than a written answer; use search_knowledge_base for a natural language answer.
query_retrieverQuery retrieverPOSTRuns a query through a Retriever, which can search several Knowledge Bases and rerank the results, and returns its outputs (text chunks or a written answer, per its response component).

Retriever, 3 operations

delete_retrieverDelete retrieverDELETEDeletes the Retriever with this id from the organization; the answer has no body.
get_retrieverGet retrieverGETReturns one Retriever with its components (name, type, description) and the schema of the query arguments it accepts.
list_retrieversList retrieversGETLists the Retrievers in the organization (managed RAG pipelines that can query several Knowledge Bases and rerank the results), a page at a time, with their components.

Event, 2 operations

list_assistant_eventsList assistant eventsGETLists the events an Assistant run emitted, with offset paging; the ASSISTANT event of type FinishStructureRunEvent carries the final answer.
list_structure_run_eventsList structure run eventsGETLists the events of a Structure run: SYSTEM lifecycle events (StructureRunStarting, Running, Completed, Error) and USER events sent by the code, such as FinishStructureRunEvent carrying the output. Offset paging.

Log, 2 operations

list_structure_run_logsList structure run logsGETReturns the log lines a Structure run printed, for debugging a failed or unexpected run.
list_tool_run_logsList tool run logsGETReturns the log lines a Tool run printed, for debugging a failed activity call.

Search, 2 operations

list_knowledge_base_searchesList knowledge base searchesGETLists the past searches made against a Knowledge Base with the answer each returned, a page at a time.
search_knowledge_baseSearch knowledge basePOSTAsks a Knowledge Base a question and returns a natural language answer (result) written from the data in its Data Sources.

Activity, 1 operation

run_tool_activityRun tool activityPOSTRuns one activity of a deployed Tool (a method decorated with @activity) with the JSON body listed by get_tool_openapi and returns the activity's answer. Each call is a Tool run that spends your Griptape Cloud usage.

Asset url, 1 operation

create_asset_urlCreate asset urlPOSTReturns a temporary presigned URL, and the headers to send with it, to upload (operation PUT) or download (operation GET) one asset of a Data Lake Bucket from outside Griptape Cloud.

Assistant run, 1 operation

get_assistant_runGet assistant runGETReturns one Assistant run: status, the message sent, the reply in output once the run has finished, the thread id and the resources used. Holds end-user conversation content.

Knowledge base query, 1 operation

get_knowledge_base_queryGet knowledge base queryGETReturns one past Knowledge Base query with the text asked and the entries it returned.

Knowledge base search, 1 operation

get_knowledge_base_searchGet knowledge base searchGETReturns one past Knowledge Base search with the question asked and the natural language answer returned.

Openapi, 1 operation

get_tool_openapiGet tool openapiGETReturns the OpenAPI document Griptape Cloud generates for a deployed Tool: one route per activity with the JSON body it takes. Read it before calling run_tool_activity.

Structure run, 1 operation

get_structure_runGet structure runGETReturns one Structure run: status (QUEUED, STARTING, RUNNING, SUCCEEDED, FAILED, ERROR or CANCELLED), arguments, output, status_detail and start and end times.

Tool run, 1 operation

get_tool_runGet tool runGETReturns one Tool run (one activity call): its input and output, status, status_detail and start and end times.

Use Griptape Cloud from Claude Code, Cursor or Codex

Connect Griptape Cloud in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Griptape Cloud 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 Griptape Cloud

Create your workspace
Sign up free. Nothing to install.
Connect Griptape Cloud once
Add your Griptape Cloud 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.