Frameworks · ESRS & custom analysis

ESRS analysis with Open Sustainability Analyst — customize frameworks on Render

Question sets define what Open Sustainability Analyst checks in a sustainability report. Edit a YAML file in your GitHub fork for custom ESRS or TCFD analysis, then deploy on Render or Streamlit. Render guide · Streamlit guide

About 15–20 minutes A free GitHub account Editing text, not code

Create your own question set

Fork on GitHub, edit a YAML question set for ESRS or TCFD analysis, and deploy your fork on Render or Streamlit.

Create a GitHub account (if you need one)

Frameworks are edited on GitHub. Create a free account if you need one — you will use it to fork, commit, and deploy.
Fork the repository

Open the report-analyst repo and click Fork — or use the direct fork link. Custom question sets must live in your fork, not the original repo.

Click Fork (top right).
Copy your fork URL

On your fork, click Code → HTTPS and copy the repository URL for deploy.

URL must show your username, not climateandtech.

Open a question set to copy

In report_analyst/questionsets/ (on GitHub), open an existing file — e.g. everest_questions.yaml or tcfd_questions.yaml. Each YAML file defines one framework: name, shortcut, description, and questions with id, text, and guidelines.

GitHub breadcrumb showing report_analyst / questionsets / everest_questions.yaml — Question sets live in report_analyst/questionsets/.


                          name: TCFD Questions
shortcut: tcfd
description: Task Force on Climate-related Financial Disclosures (TCFD) question set

questions:
  - id: tcfd_1
    text: "How does the company's board oversee climate-related risks and opportunities?"
    guidelines: |
      - Focus on board-specific oversight activities
      - Look for details about processes and frequency of oversight
  - id: tcfd_2
    text: "What is the role of management in assessing and managing climate-related risks?"
    guidelines: |
      - Focus on direct duties related to climate issues
      - Look for specific management positions or committees

Create your new question set

Use Copy raw file, then Add file → Create new file in the same folder (e.g. my_framework_questions.yaml) and paste as your starting point.

Copy raw file, then create a new .yaml in questionsets/.

Name your new file in report_analyst/questionsets/.

Edit your questions

Update name, shortcut, description, and the questions list. One clear question per row; use guidelines to steer the AI.


                          name: My Custom Framework
shortcut: mycustom
description: My own question set for analysing sustainability reports

questions:
  - id: mycustom_1
    text: "Does the company disclose a net-zero target with a target year?"
    guidelines: |
      - Look for an explicit net-zero or carbon-neutral commitment
      - Note the target year and whether it covers Scopes 1, 2 and 3
      - Avoid counting vague ambitions without a date
  - id: mycustom_2
    text: "What interim emission-reduction milestones are reported?"
    guidelines: |
      - Look for percentage reductions and baseline years
      - Note whether the milestones are science-based

Commit and deploy your fork

Click Commit changes on GitHub, then deploy your fork — not climateandtech/report-analyst. Follow the Render setup guide or Streamlit guide (deploy fork), using your fork URL.

Valid YAML files in questionsets/ appear automatically after deploy.

Tips for writing good questions

Draft with ChatGPT or Claude

Paste an existing question set and your requirements to get a first YAML draft. Test on real reports and iterate.

One idea per question

Keep each question focused on a single, answerable point.

Use guidelines to stay grounded

In guidelines, tell the AI what evidence to look for and what to avoid.

Start from the closest framework

Adapt TCFD or another built-in set instead of writing everything new.

Troubleshooting

My framework does not appear in the app

Make sure the file is inside report_analyst/questionsets/, has a .yaml extension and valid content, and that you redeployed from your fork with the latest commit.

I get a YAML error on load

Check the indentation (2 spaces, no tabs), keep quotes around each text, and use the | block style for guidelines.

Streamlit says “This file does not exist”

You are probably deploying the wrong repository (not your fork) or the main file path is incorrect. Use your fork URL and set the path to report_analyst/streamlit_app.py.

My edits are not showing

Confirm Render or Streamlit is deploying your fork (not the original repo) and that it picked up your latest commit — trigger a redeploy if needed.

Need a hand?

Tailoring ESRS question sets or need a managed setup? Use Setup or contact us.

Setup Contact us Deploy fork on Render Deploy fork on Streamlit