Data & Analytics

Querypanel

by CSABA IVÁNCZA

Overview

QueryPanel lets an AI agent configure analytics for a SaaS product without bypassing workspace permissions. Connect data sources, inspect schema context, generate and execute SQL, train business knowledge, and create charts or dashboards through one OAuth-protected MCP server. The workspace identity comes from the validated OAuth token, so tool arguments cannot switch organizations. Database credentials and private SDK keys are not returned by read tools.

Tools

add_chart_to_dashboard

ChatGPT

Activate a saved chart so it appears in the organization's active charts list. This makes the chart visible and queryable in embedded QueryPanel dashboards. The chart must already be saved (use save_chart first).

add_member

ChatGPT

Add or invite a member by email. Existing users join immediately; new users get an invite email and appear as pending until they accept.

annotate_schema

ChatGPT

Upsert a human annotation on a schema target (table/column). target_identifier uses the schema-aware form e.g. database:db:schema:public:table:orders.

clear_ai_provider

ChatGPT

Remove the organization's BYOK AI provider config.

create_dashboard

ChatGPT

Create a new dashboard in QueryPanel. Dashboards organize charts and data widgets. New dashboards start in 'draft' status.

create_datasource

ChatGPT

Create a datasource for this organization. For BigQuery OAuth connections, create the row then tell the user to finish Google OAuth in the admin UI.

create_key

ChatGPT

Upload / register a signing key for the workspace. Provide PEM public_key and optionally private_key (stored in Vault). Prefer RSA. A default key is provisioned at org creation; use generate_key to mint a new pair when you need the private PEM returned once.

delete_chart

ChatGPT

Delete a saved chart by ID.

delete_dashboard

ChatGPT

Delete a dashboard by ID. This action cannot be undone.

delete_datasource

ChatGPT

Delete a datasource by ID.

delete_key

ChatGPT

Delete a workspace signing key.

delete_knowledge_chunk

ChatGPT

Delete a gold_sql or glossary knowledge chunk by ID.

delete_schema_snapshot

ChatGPT

Remove a synced schema snapshot and its table/column embeddings. Gold SQL and glossary chunks are typically preserved. Pass table_schema_id from list_schema_snapshots.

delete_session

ChatGPT

Delete a query session and all its turn history by ID.

execute_sql

ChatGPT

Execute SQL against the organization's datasource. Pass sql and params from generate_sql. Returns rows, fields, rowCount, and a resultId.

generate_key

ChatGPT

Generate an RSA-2048 keypair for this workspace, store the public key and Vault private key, and return the private PEM once so you can write QUERYPANEL_PRIVATE_KEY to the customer .env. Call when init_sdk needs a fresh key the integrator can copy, or when list_keys shows keys but the private PEM is unknown (existing keys are never re-exported). Prefer this over create_key unless the user already has PEMs to upload.

generate_sql

ChatGPT

Generate SQL for a natural-language question using schema chunks from search_schema. Pass context_chunks through unchanged from search_schema output. Returns sql, params, dialect, database, rationale, and queryId. Pass queryId to mark_query_executed after the SQL runs outside this tool.

get_ai_provider

ChatGPT

Get the organization's BYOK AI provider status (OpenAI key configured or not).

get_chart

ChatGPT

Retrieve a saved chart by ID, including its SQL, visualization spec, and live data.

get_dashboard

ChatGPT

Retrieve a dashboard by ID, including its content and widget configuration.

get_datasource

ChatGPT

Get a datasource by ID (secrets are never returned).

get_organization

ChatGPT

Get organization details by ID.

get_session

ChatGPT

Retrieve a query session by ID, optionally including its full turn history (each question asked and the SQL generated).

get_sql_quality

ChatGPT

SQL quality rates for this workspace over the last 7 or 30 days. Returns outcome counts, fail rate, stuck drafts, gold-hit rate, latency percentiles, and pipeline mix.

init_sdk

ChatGPT

Bootstrap QueryPanel SDK for this workspace: workspace_id, readiness, env, paste-ready files. mode=headless → node-sdk qp.ask (zero-trust); mode=headful → createJwt + QuerypanelEmbedded; mode=both → both. If ensure_key and no active key, generates + Vault-saves a key and returns private_key once. Write response.files into the customer repo; never commit private keys.

list_audit_logs

ChatGPT

List query audit sessions for the organization (NL→SQL history). For BlockNote dashboard editing use the admin UI; MCP dashboards use api chart/dashboard tools.

list_charts

ChatGPT

List saved charts in QueryPanel, with optional filtering and pagination.

list_dashboard_charts

ChatGPT

List all active charts for the organization, with pagination. Active charts are the charts currently visible in embedded dashboards.

list_dashboards

ChatGPT

List dashboards in QueryPanel with optional status filtering and pagination.

list_datasources

ChatGPT

List the datasources available to this organization. Returns each datasource name, type, and id. Use the id with search_schema, generate_sql, or execute_sql.

list_keys

ChatGPT

List workspace signing keys (JWKS / public keys). Private key material is never returned — only private_key_secret_id may appear.

list_members

ChatGPT

List members of the current organization (from OAuth org / x-organization-id).

list_organizations

ChatGPT

List organizations (workspaces) the authenticated user belongs to.

list_schema_chunks

ChatGPT

Browse indexed schema chunks (table overviews, columns, gold SQL, glossary). Use filters to narrow results. For NL→SQL retrieval prefer search_schema instead.

list_schema_snapshots

ChatGPT

List scanned schema snapshots (databases that have been synced). Each entry is a table_schemas / database snapshot with table counts.

list_schema_tables

ChatGPT

List table names for a scanned database snapshot.

list_sessions

ChatGPT

List query sessions (conversation history). Each session contains a series of natural language questions and the SQL generated for them.

mark_query_executed

ChatGPT

Mark a generated query EXECUTED after its SQL has run. Pass the query id from generate_sql. A query that already shipped a chart stays SUCCESS.

remove_chart_from_dashboard

ChatGPT

Deactivate a chart by removing it from the active charts list. This hides the chart from embedded dashboards but does not delete the saved chart itself.

remove_member

ChatGPT

Remove a member from the current organization.

save_chart

ChatGPT

Save a chart to QueryPanel. Requires a title, SQL query, and a visualization spec (VizSpec from generate_visualization, or a Vega-Lite spec). Returns the saved chart with its ID.

search_schema

ChatGPT

Search the organization's indexed schema context for tables, columns, and example SQL relevant to a question. Returns ranked chunks plus the inferred primary table, database, dialect, and tenant settings. Call this FIRST, then pass its output through to generate_sql.

set_ai_provider

ChatGPT

Configure the organization's OpenAI API key (BYOK) and/or fallback-to-platform flag. Requires admin/owner role. Never echo the full key back.

set_dashboard_status

ChatGPT

Deploy or undeploy a dashboard. Set status to 'deployed' to make it live, or 'draft' to take it offline.

sync_schema

ChatGPT

Sync / re-ingest schema for a datasource into the knowledge base (embeddings). Use force_reindex=true to rebuild even when the schema hash is unchanged.

test_ai_provider

ChatGPT

Validate the org AI provider key (or a candidate api_key) against OpenAI.

test_datasource

ChatGPT

Test connectivity for a datasource.

toggle_key

ChatGPT

Activate or deactivate a workspace signing key.

train_glossary

ChatGPT

Train a business glossary term/definition into the knowledge base.

train_gold_sql

ChatGPT

Train the knowledge base with example gold SQL for a table/database. Improves NL→SQL quality for similar questions.

update_chart

ChatGPT

Update a saved chart's title, description, SQL, or visualization spec.

update_dashboard

ChatGPT

Update a dashboard's name, description, or content. content_json is a BlockNote JSON string for the dashboard layout.

update_datasource

ChatGPT

Update a datasource. Omit fields you do not want to change. Password is write-only.

update_knowledge_chunk

ChatGPT

Update a gold_sql or glossary knowledge chunk by ID. For glossary pass term+definition; for gold_sql pass sql (+ optional name/question/description).

update_member_role

ChatGPT

Update a member's role in the current organization.

update_organization

ChatGPT

Rename the current or specified organization (requires access).

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Links

Documentation ↗Privacy Policy ↗

App Stats

Tools

Prompts

ChatGPT

Platforms