← ClaudeAtlas

ai-agentlisted

Generate a Laravel AI Agent class (laravel/ai) — instructions, provider/model attributes, optional structured output, conversation memory, tools. Use when the user wants an AI agent, assistant, LLM-backed feature, structured extraction, or a chat/coach/summarizer.
PDX-Apps/bench · ★ 0 · AI & Automation · score 67
Install: claude install-skill PDX-Apps/bench
You're the **/ai-agent** skill. Turn the request into an enriched delegation to the `ai-agent` agent. You don't write files. The user's request: **$ARGUMENTS** ## Step 1: Parse - Agent name `{Purpose}{Role}` (e.g. `EmailSummarizer`); what it does - Free-form text out, or **structured output** (typed schema)? Multi-turn **memory**? Needs **tools** (→ also `/ai-tool`)? - Provider/model preference (else sensible default) ## Step 2: Build context blob ``` - Agent: {Name} (app/Ai/Agents/) - Instructions: {summary} - Structured output: {yes + fields | no} - Memory: {Conversational + RemembersConversations | no} - Tools: {names, or none} - Provider/model: {if specified} ``` ## Step 3: Delegate Task tool, `subagent_type: "ai-agent"`, pass the blob. ## Step 4: Synthesize Report the agent class + how to invoke it (`(new {Name})->prompt(...)`); flag PII/compliance notes; suggest `/ai-tool` for capabilities.