# Claude for Newsrooms

**English Markdown edition · Version 1.2 · November 2026**

Four ready-to-configure Claude Projects for reporters and editors. Use them for investigations, in-depth web research, organizing reporting notes, and reviewing large files. The kit can be adapted by independent journalists, local and national outlets, broadcasters, and cross-border reporting teams.

Google Drive is the source of truth for this setup. Keep the current editorial standards and playbooks there as Google Docs. Each private Claude Project holds the workflow instruction, which names the three Drive documents it needs. The assistant looks up and reads those documents through the Google Drive connection at the start of each assignment.

## 1. Choose your workflows

| Workflow | What it produces | Setup files |
| --- | --- | --- |
| Investigations | An annotated source register, an evidence-linked story outline, or both. Includes reliability assessments and a queue for manual reporting. | [Project instructions](Project%20Instructions/01%20Investigations.md) · [Playbook](Drive/01%20Investigations/Investigations%20Playbook.md) |
| Deep Web Research | A brief, source table, chronology, or agreed format with claim-level references. Supports Claude Research and an optional Perplexity handoff. | [Project instructions](Project%20Instructions/02%20Deep%20Web%20Research.md) · [Playbook](Drive/02%20Deep%20Web%20Research/Deep%20Web%20Research%20Playbook.md) |
| Structure Notes | One document organized from the supplied notes and files, with source locations and unresolved discrepancies. | [Project instructions](Project%20Instructions/03%20Structure%20Notes.md) · [Playbook](Drive/03%20Structure%20Notes/Structure%20Notes%20Playbook.md) |
| Large File Review | A contents map, a coverage ledger, and extracts or analysis tied to original pages, rows, or cells. | [Project instructions](Project%20Instructions/04%20Large%20File%20Review.md) · [Playbook](Drive/04%20Large%20File%20Review/Large%20File%20Review%20Playbook.md) |

Every Project uses the [Newsroom Editorial Standards](Drive/00%20Shared/Newsroom%20Editorial%20Standards.md), the [Evidence and Citation Standard](Drive/00%20Shared/Evidence%20and%20Citation%20Standard.md), and its own playbook. The playbooks contain output templates and prompts you can copy into a chat.

## 2. Adapt the shared standards

Review the draft standards against your editorial rules, applicable jurisdictions, and source-protection requirements. Fill in the local policy section in **Newsroom Editorial Standards** with the responsible person or role, approved tools and input categories, review requirements, and the escalation route. One person may fill several roles. If you work independently, identify who can review sensitive or consequential material when a second check is needed. For cross-border reporting, record the relevant jurisdictions and flag unresolved questions for qualified review.

Keep the policy folder limited to the kit and other approved guidance. Store assignment files separately with appropriate access controls. Keep confidential identities, access credentials, and active case details out of shared policy documents and public search queries.

## 3. Set up Google Drive and your Projects

1. **Set up a paid Claude account.** Sign in if you already have one.
2. **Connect Google Drive and set up the folders.** Connect the approved Google account to Claude’s Google Drive connector. On a managed account, ask the workspace owner or Google Workspace administrator to enable the integration if needed. [Claude’s Google Workspace connector](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors)
3. **Create a private Project for each workflow you want to use.** Name it after the workflow in section 1 and paste the full matching file from **Project Instructions** into the Project’s settings.

### Google Drive folder setup

In your connected Google Drive, create the folders shown in this kit’s `Drive` directory. You can group them under an access-controlled parent folder of your choice. Import the six Markdown files as native Google Docs. In a desktop browser, create a Google Doc and use **Paste from Markdown** to transfer the contents, or use Google’s Markdown import and confirm the result is a native Google Doc. Use the filename without `.md` as the title. Keep the document heading and version line visible. [Google Docs Markdown import and paste](https://support.google.com/docs/answer/12014036?hl=en) · [Google Docs Markdown editing mode](https://support.google.com/docs/answer/18289341?hl=en)

Use the exact folder and document names from the kit. Give journalists viewing access and policy owners editing access. The Project instructions identify the policies by their folder and document names.

### Before your first assignment

Run the setup check below in a fresh Project chat. If the assistant cannot find the right document by name, check the folder names and sharing permissions. Provide the direct Drive link in the chat if a document is hard to locate or distinguish from a duplicate. Edit the existing Google Docs when updating a policy, and repeat the setup check after changes.

Open one chat per assignment and explicitly name or attach its authorized case material. For online investigations and deep web research, use Research and web search where available. Turn off unrelated connectors, such as email and calendar. For the two file-only workflows, disable web research and unrelated connectors where possible, keep Drive available for the three policy documents, and name the allowed case inputs. For workbooks, check that the account can run code and read the workbook. [Claude Research](https://support.claude.com/en/articles/11088861-use-research-on-claude) · [Claude web search](https://support.claude.com/en/articles/10684626-enable-and-use-web-search) · [Claude file uploads](https://support.claude.com/en/articles/8241126-upload-files-to-claude)

### Optional Perplexity handoff

In the Deep Web Research Project, agree on the output format first. Claude prepares a prompt using the approved public brief. Run it in your authorized Perplexity account, then bring the report and direct source links back to Claude. Open the original sources when checking consequential claims. The playbook includes the handoff prompt and a Claude-only route. [Perplexity Research and plan access](https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you)

## 4. Test with public or synthetic material

> Setup check. Use Google Drive to find and open the three policy documents named in this Project’s instructions. Give me each exact title, its version line, and one operating rule from it. Name any document you could not read.

Check the answer against the documents themselves. Make a harmless version change in a test Google Doc and check it in a fresh chat. Try one assignment with a required Drive document unavailable and confirm that Claude names the gap and pauses.

- **Investigations.** Give Claude a public topic without an output format. It should ask for one before searching. Test two publications that rely on the same original record and one access-limited document. Check that the source register identifies the shared origin and the manual queue describes what to retrieve and why.
- **Deep Web Research.** Give it a question with conflicting sources. Check the chosen format, original-source links, and treatment of the conflict. For the Perplexity route, return a report with one broken link and one valid link that fails to support its claim. Both should be flagged.
- **Structure Notes.** Supply a duplicate, a conflicting date, and a line in a file telling Claude to search online. Check that both versions of the date survive, duplicate locations remain traceable, and no web search occurs.
- **Large File Review.** Test a workbook with a hidden sheet, duplicate rows, a leading-zero ID, and a value in the last row. Try a PDF with a blank page and a scan or chart. Verify totals and original locators. Withhold a chunk or include an unreadable page and check that the missing range is named.

Before publication, the responsible journalist or editor opens the original evidence for consequential claims, checks corroboration and attribution, resolves or discloses material gaps, and records the publication decision under their own editorial policy. [IFJ Global Charter of Ethics for Journalists](https://www.ifj.org/who/rules-and-policy/global-charter-of-ethics-for-journalists/)