Works with the AI Agents you already use
Looker Studio Dashboard Generator (Text to Dashboard on BigQuery)
Turn "executive sales dashboard with revenue trend and store ranking" into a real Looker Studio report on BigQuery, without the LLM ever inventing a field.
$15
Looker Studio Dashboard Generator (Text to Dashboard on BigQuery)
Example session with this skill installed
Our BI team wants a manager to type "show me revenue and margin by month and the top stores" and get a Looker Studio dashboard over acme-prod.sales.transactions (transaction_date, store_name, category_name, sales_amount, gross_margin, customer_email). We use Gemini. Write the design in DESIGN.md and the semantic model in semantic_model.yaml.
- Read your context and instructions
- Compiled the looker studio dashboard
DESIGN.md: Gemini fills a DashboardSpec via structured output using only semantic names (date, store, category, revenue, margin); the compiler rejects unknown or sensitive fields and a time series on a non-time dimension, selects the executive_sales master template, and returns a Linking API URL (c.reportId, ds.ds0.connector=bigQuery, type TABLE, project/dataset/table). Template built on a view with stable column names. Allow-list acme-prod.sales. semantic_model.yaml: revenue → sales_amount SUM, margin → gross_margin SUM, date → transaction_date (time_dimension), store → store_name, category → category_name, customer_email sensitive: true.
Connects securely to your tools. The creator never sees your data.
About this skill
The tempting design (ask the model to emit a Looker Studio report) does not work: the report format is internal and undocumented. The design that works keeps the LLM on intent and field selection and does everything else deterministically.
What your agent builds with it
- A semantic model per BigQuery table: business names → physical columns, roles, aggregations,
sensitiveflags. It is the only vocabulary the LLM sees; PII never reaches the model. - A dashboard-spec JSON contract you own (JSON Schema included) filled with structured output, plus a one-round repair loop fed by the validator's errors.
- A deterministic compiler (
dashboard_compiler.py): rejects unknown or sensitive fields, time series on non-time dimensions, unsupported charts and too many components; selects a template; maps fields; builds the Looker Studio Linking API URL for a BigQuery TABLE source. - Owned master templates and the one pitfall that breaks most attempts: charts bind to fields by name, so templates must sit on stable column names.
- Governance: table allow-lists, least-privilege identity, structured statuses for every exit path, audit logging.
- A conversational analyst agent (ADK) that explores data with guarded SQL and hands off "chart this" to the same compiler.
- A test plan: compiler unit tests and a 20–50 prompt set that measures fabricated-field rate.
Included: SKILL.md, prompt and agent reference, dashboard_compiler.py (standard library; PyYAML optional), spec JSON schema, example semantic model and template registry, and an eval case.
Part of the Production AI Agents series. Pair with Text-to-SQL Data Agent for the analyst's SQL guard and BigQuery Data Catalog for the semantic layer.
Uses master templates rather than fully generative layouts (by design). The Linking API flow supports BigQuery TABLE sources; custom-query sources are out of scope. Verify Linking API parameter names against Google's current docs before production.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
Requires
Permissions
File Scopes
dashboard_compiler.py makes no network calls; it only builds a URL that opens in the user's browser.