opentargets-cli

Agent-ready CLI for querying the Open Targets Platform GraphQL API.

Install the ot CLI, start Claude Code or Codex, and ask Open Targets questions in plain English. This repo ships the CLI and agent instructions; the agent builds the query, while ot handles transport, ID resolution, schema inspection, JSON output, and provenance.

How it works

flowchart LR
    USER([You])
    AGENT[Claude Code / Codex]
    SKILL[(Repo instructions<br/>or installed skill)]
    CLI[ot CLI]
    API[(Open Targets<br/>GraphQL API)]
    OUT[Chat answer<br/>with IDs + provenance]

    USER -->|chat| AGENT
    USER -->|shell| CLI
    SKILL -.guides.-> AGENT
    AGENT -->|invokes| CLI
    CLI <--> API
    CLI --> OUT

Loading

Quick start

Prerequisite: install uv.

From package

uv tool install git+https://github.com/nickzren/opentargets-cli.git
ot install-skills
ot doctor
codex                              # or: claude

From repo

git clone https://github.com/nickzren/opentargets-cli.git
cd opentargets-cli
./scripts/install.sh               # install the CLI, skill(s), and run checks
codex                              # or: claude

Then ask:

What diseases is BRCA1 associated with?

Show drugs linked to rheumatoid arthritis.

What evidence supports EGFR in lung adenocarcinoma?

For repo-local use, start Codex or Claude Code from this directory. AGENTS.md and CLAUDE.md tell the agent how to use ot.

For chat-anywhere use, run:

ot install-skills

ot install-skills installs detected agent skill(s). If a skill already exists, it is moved to a timestamped backup before installing the new copy. Use ot install-skills --agent all to install both Claude Code and Codex skills explicitly.

Restart the agent session after installing or updating the skill.

CLI usage

The same tool works as a plain shell CLI:

ot tools
ot doctor
ot install-skills
ot meta --no-downloads
ot resolve "BRCA1" --entity target
ot schema --category targets --category associations
ot gql --query 'query { meta { name } }'
ot gql --query-file query.graphql --variables-list vars.ndjson --key-field ensemblId

ot exposes seven agent-facing primitives (meta, tools, describe, resolve, schema, type, gql) plus two setup commands (doctor, install-skills). Output is JSON by default and includes status plus request context; live API commands also include Open Targets API/data-release provenance.

Run ot --help or ot describe <command> for the full command contract.

Agent Rules

Agent-specific workflow rules live in AGENTS.md, CLAUDE.md, and skills/opentargets-cli/SKILL.md.

Validation

python3 -m compileall src
python -m pytest tests
python -m pytest tests -m live
python -m pytest tests/skill -m 'live and skill'

Offline tests run by default. Live and skill-pattern tests are opt-in checks against the Open Targets API.