Skip to content

Set up in Claude Code

Install bharat-judgements where your local Claude Code session can execute it. The bundled skill supplies instructions; the package supplies the actual queries.

Docker MCP connection

A local Docker MCP server is now included. Follow Docker MCP setup to connect supported local assistants and verify their tools before researching cases.

01 / Install the tool

You need Python 3.11 or newer. In a terminal, run:

python -m pip install "bharat-judgements[cli,ocr,archive]"
bharat-judgements --help
bharat-judgements version

cli supplies the command-line dependencies. ocr supplies the default automatic CAPTCHA solver. archive supplies historical queries. OCR dependency support varies by Python version and platform; see installation if the install fails.

02 / Add the skill to your working folder

Open a terminal in the folder you use for research, then run:

bharat-judgements install-skills

The command copies the bundled files into .claude/skills/bharat-judgements/ under the current folder. If that folder already exists, its bundled files are overwritten. Keep your own amendments separately before reinstalling.

Local Claude Code discovers project skills under .claude/skills/, as described in Anthropic's skills documentation. Start Claude Code in this folder and confirm that the skill is available.

03 / Check execution before querying a matter

Ask Claude to run bharat-judgements version and bharat-judgements --json courts --type hc. Both checks are offline. The output verifies that this session can execute the installed command; it does not verify live portal connectivity.

Example research prompt:

Use bharat-judgements to find Supreme Court judgments by Chandrachud from 2018 through 2024. Return up to ten records with their dates and identifiers. State the source and any missing fields.

For a live record, the session also needs outbound network access. Follow your normal tool permissions when approving execution. A skill does not grant those permissions.

When setup fails

  • Command missing: check that the terminal and Claude use the same Python environment and that its scripts directory is on PATH.
  • Skill missing: confirm the working folder contains .claude/skills/bharat-judgements/SKILL.md.
  • CAPTCHA fails: verify the ocr extra installed successfully. Automatic solving can still exhaust its retry budget.
  • Archive unavailable: verify the archive extra and network access. Empty results and connection errors mean different things.

Cowork and hosted sessions

Do not assume a local project folder gives Cowork or a hosted session the same executable access. Account-enabled skills and execution environments have separate setup. Follow the Desktop and Cowork guide, or use the local CLI workflow.

Next: try a prompt or inspect the available commands.