Skills and Capability

Capability: tool layer

A Capability is a tool bundle the LLM can call. Each cap_* module implements a feature (Lua, IM, files, …) and registers a Capability Group with claw_cap.

On every LLM request, claw_cap_tools_provider injects only tools from groups that are visible for the session—hidden groups do not exist in context. That avoids dumping unrelated tools.

Skills: knowledge layer

A Skill is a Markdown file describing how to use a feature, workflows, or behavioral contracts. claw_skill injects via two context providers:

Provider	Injected content	Always on
`claw_skill_skills_list_provider`	All Skill ids, summaries	yes
`claw_skill_active_skill_docs_provider`	Full `.md` for activated Skills	only when active

Before activation the LLM only sees summaries; activation loads the full doc. Activation is per session and persisted on disk across reboots.

Relationship and differences

Capabilities expose tool signatures without prose guides; Skills carry guides without executable tools—they complement each other:

	Capability	Skill
Nature	Tools (functions)	Knowledge (context)
Carrier	C code (`cap_*`)	Markdown files
Role	Lets the LLM do	Teaches the LLM how
Lifecycle	Registered at boot; visibility per group	Activate/deactivate on demand; persisted

They link through cap_groups in skills_list.json:

{
  "id": "cap_lua_run",
  "cap_groups": ["cap_lua"]   // opening this Skill also opens the cap_lua tool group
}

Activating a Skill does two things:

Injects the .md into the next request (knowledge).
Makes the listed capability groups visible for the session (tools).

Deactivating removes both to keep context tight.

claw_skill Skills layout, skills_list.json, and activation plumbing

cap_skill Console skill commands and activate_skill / deactivate_skill tools

Skills guide Manual and Agent workflows plus file format examples