# OpenClaw Input Example

## Project

- Project name: OpenClaw
- Domain: embodied AI / robotics
- Goal: turn a manipulation system project into a paper-ready evidence pack
- Target writing tasks: related work, contribution framing, experiment planning, draft support

## User-provided project context

- OpenClaw combines perception, planning, and action components for manipulation tasks
- The team wants to write a paper without inventing claims or overstating results
- The team expects the workflow to search real prior work, datasets, benchmarks, and public examples
- The team may have internal results, but unsupported quantitative claims should not appear in draft text

## Constraints

- Do not fabricate experiments, numbers, citations, baselines, or benchmark wins
- Use primary sources whenever possible
- Separate verified facts from inference
- If a strong claim cannot be supported, mark it as a gap

## Desired output

1. Evidence brief
2. Key sources and why they matter
3. Claim-to-evidence map
4. Evidence gaps
5. Safe writing notes for later drafting