A fully AI-driven repository of mid-level robotics skill adapters.
Level definitions:
- High-level: cross-library strategy and planning guidance.
- Mid-level: semantic "how to use" guidance for specific libraries.
- Low-level: implementation details such as shell commands or scripts.
Skills are organized by robotics library, with one or more prefix-based files per library.
skills/
lib_name/
lib-skills.md
Use AI_PROMPT_TEMPLATE.md to generate consistent, auditable contributions.
Prompt e.g.
Please help me contribute to this repo based on AI_CONTRIBUTING.md, with curobo_skills.md on Desktop
Please help me contribute to this repo <path to repo> based on its AI_CONTRIBUTING.md, with our previous conversion about using curobo.The template supports:
- website/documentation research,
- local repository knowledge extraction,
- experiment result summarization,
- update of existing skill files or creation of new ones.
You should contribute when you can provide one of the following:
- A reusable robotics skill update for an existing library file.
- A new prefix-based skill file for a library that does not have one yet.
- A correction or improvement based on verified docs, code, or experiments.
- Keep each PR focused on one library folder.
- Update only prefix-based skill files in skills//.
- Include a short evidence note in the PR for why the update is valid.
You only need to provide three things to your AI system:
- Ask the AI to read AI_PROMPT_TEMPLATE.md first.
- Tell the AI which library you want to contribute to (for example: ros2 or moveit).
- Tell the AI which evidence source to use:
- web search
- local repository analysis
- a summarized .md prepared by another agent
Then ask the AI to update or create exactly one target file in skills//-skills.md using the required skill structure defined in AI_PROMPT_TEMPLATE.md.
- Contributing guide: CONTRIBUTING.md
- AI contribution rules: AI_CONTRIBUTING.md
- AI prompt template: AI_PROMPT_TEMPLATE.md
- Blind test protocol and latest result: BLIND_TEST_PROTOCOL.md
- Security policy: SECURITY.md