ai-agent-architecture

Plugin ai-agent-architecture from marketplace joycai-agent-skill in Joycai/Joycai-agent-skill. Commands read from the repo's manifest — 6 days ago.

Skills (5)

• agent-runtime-architecture

  Design, implement, or review agent runtimes: tool-loop lifecycle, cancellation and recovery, execution permissions, approval queues, delegation, context compaction, and session persistence. Use for stuck approvals, missing subagent results, or broken session state. Endpoint wire formats belong to ll

• ai-media-api-integration

  Integrate or debug image, video, and speech-transcription APIs: request formats, uploads, async jobs, polling, cancellation, downloads, and timestamps. Use for API engineering, not media creation with existing tools. version: "1.1.9" updated: "2026-11-06

• ai-usage-accounting

  Design or review usage metering and cost accounting for LLM, image, and video APIs: normalized usage, price snapshots, reported costs, async settlement, and usage dashboards. Use for cost tracking and reconciliation, not subscription payments, invoicing, or collecting money from customers. version:

• llm-api-integration

  Design, implement, audit, or debug conversational LLM API integrations and check platform × protocol × model compatibility. Use for request/response formats, streaming, reasoning, structured output, tools, and endpoint failures. Media generation and transcription APIs use ai-media-api-integration; m

• llm-provider-config

  Design or migrate LLM provider configuration: credentials, endpoints, protocol routes, model settings, and capability profiles. Use for multi-protocol credentials, per-route settings, duplicate-provider merging, or configuration UI. Protocol wire behavior and actual vendor capabilities belong to the

Similar plugins