Task summary
Introduce a lightweight “installable skills” mechanism so skills can be packaged and loaded dynamically instead of being tightly coupled to the core codebase.
The goal is to make OmniClaw easier to extend by allowing contributors to add reusable capabilities without modifying core logic.
This task would explore adding a small framework that allows skills to be discovered and loaded from a dedicated folder structure and optionally enabled or disabled depending on the agent or workspace.
Area
SDK
Context
A modular “installable skills” approach could make the project easier to extend and maintain, especially as more contributors start building custom capabilities.
Possible benefits:
• contributors can add new skills without modifying core files
• easier experimentation with new agent capabilities
• skills can be reused across different agents or environments
• users can enable or disable skills depending on their workflow
Example skills that could benefit from this approach:
• GitHub PR review assistant
• repository issue triage
• scheduled summaries or reports
• wallet/payment monitoring agents
The goal would be to keep the first version very small and backwards-compatible with the current architecture.
Acceptance criteria
• OmniClaw can discover skills from a dedicated skills directory at startup.
• Each skill has a small manifest file containing metadata such as id, name, description, and entry configuration.
• Invalid or malformed skill manifests do not crash the system and are skipped with a clear log message.
• Skills can be enabled or disabled through configuration without modifying core code.
• At least one example skill is added to demonstrate how contributors can create new skills.
• Documentation explains the expected folder structure and how contributors can add new skills.
References
N/A
Suggested difficulty
Good first issue
Task summary
Introduce a lightweight “installable skills” mechanism so skills can be packaged and loaded dynamically instead of being tightly coupled to the core codebase.
The goal is to make OmniClaw easier to extend by allowing contributors to add reusable capabilities without modifying core logic.
This task would explore adding a small framework that allows skills to be discovered and loaded from a dedicated folder structure and optionally enabled or disabled depending on the agent or workspace.
Area
SDK
Context
A modular “installable skills” approach could make the project easier to extend and maintain, especially as more contributors start building custom capabilities.
Possible benefits:
• contributors can add new skills without modifying core files
• easier experimentation with new agent capabilities
• skills can be reused across different agents or environments
• users can enable or disable skills depending on their workflow
Example skills that could benefit from this approach:
• GitHub PR review assistant
• repository issue triage
• scheduled summaries or reports
• wallet/payment monitoring agents
The goal would be to keep the first version very small and backwards-compatible with the current architecture.
Acceptance criteria
• OmniClaw can discover skills from a dedicated skills directory at startup.
• Each skill has a small manifest file containing metadata such as id, name, description, and entry configuration.
• Invalid or malformed skill manifests do not crash the system and are skipped with a clear log message.
• Skills can be enabled or disabled through configuration without modifying core code.
• At least one example skill is added to demonstrate how contributors can create new skills.
• Documentation explains the expected folder structure and how contributors can add new skills.
References
N/A
Suggested difficulty
Good first issue