Move exported work to Querio

Already unshackled your BI tool? Open the exported repository in your coding agent. It can use the queries, report definitions and business context to rebuild your work in Querio, starting with one report you can check.

If you haven't exported yet, choose your BI tool.

1. Connect Querio CLI

Install Querio CLI, then sign in:

querio login
querio me
querio workspace list
querio workspace use WORKSPACE_ID

Replace WORKSPACE_ID with your team's workspace ID. Already a customer? Use your existing account and workspace. New to Querio? You can sign up during browser login, then follow Get started with CLI to create your workspace.

Use the same database or warehouse the original report reads. Reuse your connected Querio datasource, or connect it first. The agent uses your permissions; it needs access to that data and permission to create and run notebooks.

2. Give your agent the source material

Have these ready in the repository:

Paths can differ if you customized unshackle.toml. Review the files you give your agent; raw exports can contain data or identities you don't want to share.

Record the baseline's dates, filters, timezone, currency and access scope. For example: January revenue by region, in USD, excluding refunded orders. Pin relative dates so the comparison doesn't change while you work.

3. Ask your agent to rebuild the report

Paste this into Claude Code, Codex or another coding agent with terminal access. Replace the three uppercase placeholders:

Move REPORT from this unshackle repository to Querio workspace
WORKSPACE_ID using Querio CLI and datasource DATASOURCE_SLUG.

Read migration/report.md, the report's exported files and unresolved
TODOs. Read the repository's migration guides if present, then run:
querio code skill
querio code skill --reference notebook-authoring

Inspect existing Querio context before changing it. Reuse relevant
business definitions and propose any additions or conflicts.

Rebuild this report as Querio notebook cells, with a board showing
its useful outputs and controls. Query the configured Querio datasource
rather than using the export's standalone warehouse connection.
Preserve metric definitions, filters, aggregation grain and chart meaning.
Flag unsupported expressions instead of guessing.

Run the actual cells and compare rows and totals with my source result
using the same dates, filters, timezone, currency and access scope.
Ask me for missing comparison conditions before claiming a match.

Confirm saving. Return the notebook ID, a comparison of the results,
context changes and any unresolved gaps. Close only the session and
execution environment you created, after saving succeeds.

The exported Jupyter notebook is source material for the agent, not a file you can import unchanged into Querio. The CLI's bundled guides explain how to author cells and select the datasource for execution.

Add vetted definitions through Querio's context files, CLI or Git workflow. Exported policies describe source access intent; configure Querio users, roles, groups and datasource access separately.

4. Check the work in Querio

Ask the agent to show the queries, calculations, charts and comparison. Check more than the headline total: grouping, row counts and filter behavior should agree with the original report. Set any numeric tolerance before accepting differences.

unshackle verify can check exported queries against a source baseline. It does not prove that the rebuilt Querio cells match. Check and correct an answer covers inspecting the actual analysis.

Confirm that the work saved:

querio notebook get NOTEBOOK_ID --json

Open the Explore and inspect its notebook and board. Test the controls your teammates will use. If saving failed, resolve it in the existing session before closing or repeating the task.

5. Share it, then move on to the next report

Share the Explore or board with an intended reader and check that their access works. Keep the source report, repository revision, comparison conditions and remaining gaps with the handover.

Keep the original report available until the replacement passes your checks. Then use the same process for the next report, reusing the context your team has already vetted.