4 ms·
Claude Code, at least, will only load a SKILL.md file into context when it’s invoked by the user or LLM itself, i.e. in demand.
by danw1979 7mo ago
Claude Code, at least, will only load a SKILL.md file into context when it’s invoked by the user or LLM itself, i.e. in demand.
- sheept 7mo agoClaude will load the name and description of each enabled skill into context at startup[0]; the LLM needs to know what it can invoke, after all. It's negligible for a few skills, but a hundred skills will likely have some impact, e.g. deemphasizing other skills by adding noise. [0]: https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview https://platform.claude.com/docs/en/agents-and-tools/agent-s...
- danw1979 7mo agoGood point. Also, the author specifically mentions OpenClaw in the example skill frontmatter, so I'm wondering if their workflow even involves CC.
- mkobit 7mo agoYou could be thinking of the Claude tool search tool [0][1], which lets it do that on-demand loading. Skills (like another comment mentioned) have a progressive disclosure mechanism built-in. Agent Skills [2] appear to be the common format across most of the tools, but I have no idea if this actually has tool provider buy-in. [0]: https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool https://platform.claude.com/docs/en/agents-and-tools/tool-us... [1]: https://www.anthropic.com/engineering/advanced-tool-use https://www.anthropic.com/engineering/advanced-tool-use [2]: : https://agentskills.io/home https://agentskills.io/home