An autonomous AI agent that works until it finds a cure to every disease. Built on OpenClaw, it separates planning from execution and orchestrates full discovery campaigns end to end.
ClawCures is a disease-campaign orchestration system that separates
planning from execution. OpenClaw generates
strategic plans through its /v1/responses endpoint, while
refua-mcp executes typed scientific tools in a strict, auditable
workflow.
By default, ClawCures pursues a singular, ambitious objective: find cures for all diseases by prioritizing the highest-burden conditions and researching the best drug design strategies for each.
Set a disease target or let ClawCures default to its all-disease mission.
The planner generates a strict JSON tool-call plan through the OpenClaw gateway.
Policy checks and optional human review ensure the plan meets safety and scientific constraints.
Typed scientific tools run the actual folding, affinity predictions, and design workflows.
In autonomous mode, a critic evaluates results and triggers additional rounds as needed.
Execute a single planner + execution cycle. Supports custom objectives, plan files, dry-run mode, and output artifact paths.
Multi-round planner/critic loop that autonomously iterates until the campaign objective is satisfied or limits are reached.
Policy-check a local JSON plan offline without making any API calls or running scientific tools.
Rank disease programs from a JSON input file to prioritize the most impactful targets.
pip install ClawCuresClawCures runOr do a dry-run first:
ClawCures run --dry-runClawCures is the autonomous engine behind the AgentCures mission. Start a campaign, review the plan, and let AI drive discovery forward.