accept_workspace
ChatGPTAccept the finished Intro build after the user reviews its results. Same as Open workspace in Jsonify. Makes the workspace available for normal use; does not enable schedules, send anything, or upgrade its plan. Requires the workspace owner and a completed build.
browse_dataset_rows
ChatGPTRead one rich table page with search and sorting, capped at 100 rows and 96 KB.
build_workspace
ChatGPTCreate the confirmed workspace and collect its first bounded sample from the web sources in the brief. This is where row metering starts; preparation is free. Before calling, show the user the brief and the first_rows and monthly_rows estimate from get_conversation, and pass accepted_terms=true only after they agree to build and accept Jsonify terms. The review_hash binds this call to that exact brief. Reads sources and never posts to them. Repeating the call returns the same build.
cancel_work
ChatGPTInterrupt the specified queued or running work input when the user asks to stop it. This ends that input and may discard unfinished candidate work; retrying does not resume it. Does not undo completed changes or stop unrelated production runs. Requires access to the conversation; repeated cancellation is safe.
copy_dataset
ChatGPTCopy an eligible public dataset and its runnable data collection pipeline into a new free workspace in your connected organisation. Requires authenticated organisation access. Copies the entire existing snapshot without collecting fresh data or charging imported rows; 100 free collection rows remain and schedules start disabled. Returns workspace_id, dataset_id, pipeline_id and a browser URL for immediate query_dataset, browse_dataset_rows or export_dataset use. Use the source dataset public ID and a new UUID idempotency_key; reuse both on retry to return the same workspace. No setup, build_workspace or accept_workspace step is needed. Private dependencies and saved site logins are unsupported. Future run_pipeline calls use the usual estimate and confirmation flow.
create_workspace
ChatGPTCreate an empty workspace when requested. Select organisation_id from get_context when more than one is available. Reuse the same idempotency_key on retries. Does not start collection or paid work.
detect_directory_source_urls
ChatGPTAdmin: detect unique HTTPS source domains from pipeline code, configuration and page copy; does not save.
export_dataset
ChatGPTQueue an export of the full selected immutable table, or read its status using job_id. Download requires Jsonify browser sign-in.
generate_directory_sample
ChatGPTFactory platform admins only. Spend AI usage to select representative real rows for one sheet. Returns an UNSAVED draft; use save_directory_sample to publish after review.
get_context
ChatGPTShow the authenticated organisation, access and up to 50 workspaces. Page with cursor.
get_conversation
ChatGPTRead public events for this exact input. Omit cursor on the first read, then pass the returned next_cursor to read only newer events. Relay Jason's questions and answers. Closing the host does not stop work.
get_dashboard_filter_options
ChatGPTRead available choices for one registered filter on an existing workspace dashboard, optionally constrained by current filter selections. Use a filter_id returned by get_dashboard_view. Queries existing dataset values without changing saved filters, dashboards or data.
get_dashboard_view
ChatGPTRead an existing workspace dashboard's layout, registered queries, filters and dataset references, plus a temporary read-only render URL. Use query_dashboard for chart results and get_dashboard_filter_options for filter choices. Does not create or modify the dashboard or source data.
get_dataset
ChatGPTRead one exact table's schema (up to 100 columns) and version for subsequent row queries.
get_dataset_view
ChatGPTRead an existing workspace dataset's description, table row counts, selected table schema and up to 20 revisions. Omit version to use the active version; pass next_before as before to page older revisions. Use browse_dataset_rows to read records. Does not modify or export data.
get_directory_page
ChatGPTFactory platform admins only. Read the complete saved public dataset-page content and revision, identified by its pipeline ID in the connected organisation. These are the fields used by Public Directory and the public website, not the pipeline brief or workspace share link. Includes the public URL, live schedule facts, published-data update time, and latest recorded quality verdict with its assessed pipeline version. Use this before update_directory_page.
get_directory_sample
ChatGPTFactory platform admins only. Read the complete saved sample and edit revision for one published sheet. Null sample means automatic first rows. Use get_dataset to discover output sheets.
get_pipeline
ChatGPTRead the selected pipeline's active version, last run status, cadence and whether its schedule is enabled.
get_pipeline_view
ChatGPTLoad a selected pipeline's description, latest production run and bounded output previews.
get_run
ChatGPTRead the selected run's recorded status and up to 20 output row counts. No full logs or rows.
get_starter_ideas
ChatGPTLoad bounded onboarding cards for the selected workspace.
get_workspace_usage
ChatGPTRead the selected workspace's current usage, collection cap and pricing.
import_directory_sample_csv
ChatGPTFactory platform admins only. Parse UTF-8 CSV with unique current sheet column headers into an UNSAVED sample draft. Cells remain text; existing sample byte and row budgets apply. Use save_directory_sample to publish.
list_conversations
ChatGPTFind up to 50 interactive Jsonify conversations for explicit continuation; does not load histories.
list_datasets
ChatGPTFind up to 50 datasets and their active versions. Use get_dataset for a selected schema.
list_directory_pages
ChatGPTFactory platform admins only. Discover one page of public directory entries in a workspace in the connected organisation, including listing visibility, public URLs, schedule facts, data-update times and latest recorded quality verdicts. Includes hidden/noindex/index listings with publication consent and published data. Follow next_cursor to read all entries. No page bodies, dataset rows or report histories are loaded; select get_directory_page to edit one page.
list_pipelines
ChatGPTFind up to 50 pipelines using recorded summaries; no per-item detail loading.
open_pipelines
ChatGPTOpen Jsonify. One bounded page of workspace and pipeline summaries.
query_dashboard
ChatGPTRun one registered chart query. No caller-supplied SQL.
query_dataset
ChatGPTRead exact typed rows from one dataset table, with optional equality filters. Select columns; pages stay pinned to one version. Oversized results require fewer rows/columns. No SQL or credentials.
request_work
ChatGPTBuild new pipelines in the current chat workspace with workspace_id and workspaceless=false. Creating a pipeline is not creating a workspace. Use workspaceless=true only for an explicit new-workspace request or first-time onboarding. If no workspace is known, ask the user to select one. Otherwise build, change, or diagnose data pipelines, or continue an existing conversation. Jason handles authoring, testing and reviewed activation, reading the web sources involved. Jason can delete a pipeline or dataset only when the user explicitly asks; other changes are versioned. Jason may change pipelines, run them and set schedules. Jason's test runs are free; production and scheduled runs count rows against the workspace's capped monthly allowance. It sends data outside Jsonify, such as email or webhooks, only to destinations the user sets up or asks for. Returns promptly; follow the input with get_conversation. Reuse the exact key and arguments on retry.
reset_directory_sample
ChatGPTFactory platform admins only. Remove the curated sample and use automatic first rows again. Requires the current saved sample revision.
run_pipeline
ChatGPTRun an existing activated pipeline, capped at row_limit. Two steps: without confirmation this only returns the row allowance and maximum charge estimate and runs nothing. Show it to the user; after they agree, call again with the same arguments plus the returned confirmation to start the run, then inspect it with get_run. Approval binds the selected revision and retry key; if the active revision changes before confirmation, obtain a fresh estimate and approval. Reads the pipeline's web sources and never posts to them; notifications the user already configured in Jsonify may fire when it finishes. Does not author code or change schedules. Reuse the exact key and arguments on retry.
save_directory_sample
ChatGPTFactory platform admins only. Replace one sheet sample with exact supplied rows. Pass the revision from get_directory_sample; stale edits fail. Saved rows persist across dataset updates.
search_public_directory
ChatGPTFind published datasets across the public directory by title, description or dataset name. Empty query browses one page. Follow next_cursor with the same query for more results. Returns dataset IDs, descriptions, row counts and update times; no private data or dataset rows. After the user selects a result, use copy_dataset with its dataset_id and a target organisation from get_context. Copy rechecks eligibility and imports the existing snapshot and pipeline into a new workspace without collecting or charging rows.
set_directory_page_visibility
ChatGPTFactory platform admins only. Curate the landing website: unlisted keeps the page accessible by direct URL but out of landing lists and search; noindex lists it on landing without allowing search indexing; index lists it and permits indexing. Read get_directory_page first and pass its content revision. Index requires a public title, meta description and overview. Does not change participation in Factory's in-product Public Directory or override the owner's publication consent. Returns the saved content and landing setting.
set_directory_sample_sort
ChatGPTFactory platform admins only. Set the one shared default order used by sheet settings, sample preview and landing. Empty column clears sorting.
set_pipeline_directory_visibility
ChatGPTSet whether a pipeline participates in Factory's Public Directory, only when the user requests it. On publishes its details, steps and all results including future updates once data is ready; off removes public access; inherit follows the workspace default. Published details and samples may be copied to public GitHub, where prior copies can remain. New pipelines default to unlisted on the landing website. This does not change admin landing curation, enable build-conversation sharing, collect data or change schedules. Requires normal access to the connected organisation and workspace, not platform admin.
start_preparation
ChatGPTStart a private Jason preparation with OAuth or an anonymous WorkOS agent token. Returns a private preparation_token and session_id. The token is a random possession proof scoped to this onboarding conversation, not a password, API key or OAuth access/refresh token; it is not signed. It allows access to the preparation and transfer into browser signup, expires after 24 hours if unclaimed, and requires the workspace owner after creation. MCP calls still require an authenticated access token; anonymous agent tokens allow preparation only. Retain both for request_work(workspaceless=true), get_conversation and cancel_work. No collection or workspace is created. Do not expose the token except in the returned trusted signup link.
update_directory_page
ChatGPTFactory platform admins only. Update only the supplied public page fields and return the complete saved content and new revision. Omitted fields remain unchanged; supplied lists replace that entire list, and empty strings/lists explicitly clear fields. Read get_directory_page first and pass its revision; stale edits are rejected. Changes affect Public Directory and the public website immediately on their next uncached read. Use the user's keyword research and editorial rules to write factual copy before calling. This does not run Jason, collect data, change indexing/publication consent, or edit workspace shares. License changes require an explicit user decision. After an uncertain response, read the page to check whether the edit saved before retrying.
update_directory_pipeline_identity
ChatGPTFactory platform admins only. Change the REAL pipeline name and icon (empty icon clears it), including its slug. This is separate from the public-title override. Read get_directory_page first.
update_workspace_row_limit
ChatGPTSave the cap after the user confirms in the usage panel. May pause collection. Changes no payment method and creates no charge. Existing billing rules apply.
accept_workspace
Claudebrowse_dataset_rows
Claudebuild_workspace
Claudecancel_work
Claudecopy_dataset
Claudecreate_workspace
Claudeexport_dataset
Claudeget_context
Claudeget_conversation
Claudeget_dashboard_filter_options
Claudeget_dashboard_view
Claudeget_dataset
Claudeget_dataset_view
Claudeget_pipeline
Claudeget_pipeline_view
Claudeget_run
Claudeget_starter_ideas
Claudeget_workspace_usage
Claudelist_conversations
Claudelist_datasets
Claudelist_pipelines
Claudeopen_pipelines
Claudequery_dashboard
Claudequery_dataset
Clauderequest_work
Clauderun_pipeline
Claudesearch_public_directory
Claudeset_pipeline_directory_visibility
Claudestart_preparation
Claudeupdate_workspace_row_limit
Claude