Atlas › 01 Agent Skills for Browser Automation › Skill Lifecycle: From Installation to Execution☰ Read as one page
Skill Lifecycle: From Installation to Execution
2.1Phase 1: InstallationSkills are installed from Git repositories using the skills CLI, or generated by the tool they wrap:2.2Phase 2: DiscoveryOn every conversation turn, the system:2.3Phase 3: SelectionWhen a user message arrives, Claude's language model processes: - The user's message - Conversation history - All available tools…2.4Phase 4: InvocationWhen Claude decides to use a skill, it calls the Skill tool:2.5Phase 5: ExecutionWith the SKILL.md instructions in context and tools available, the agent proceeds to execute. For a browser skill, this typically means:2.6Phase 6: CompletionWhen the agent finishes the task (or the user interrupts), the skill's effects unwind: - Elevated permissions revert to the session's…2.7Lifecycle Diagram2.8Key TakeawayThe elegance of the skill system is that it adds zero new infrastructure. No servers, no APIs, no protocols. Just a markdown file that gets…