AI Application Architecture
Designs retrieval pipelines from the corpus: chunking, embeddings, retrieval strategy and how answers cite sources.
Adrian builds LLM features and agents: retrieval pipelines, prompts under version control, evaluations, tool servers and cost controls.
A complete skills pack configured for this role. Source files and licenses are included in the download.
Designs retrieval pipelines from the corpus: chunking, embeddings, retrieval strategy and how answers cite sources.
Chooses the orchestration pattern for multi-agent systems and defines tools, handoffs and stopping rules.
Writes prompts with explicit inputs, constraints and failure cases, and tests them against edge inputs.
Designs retrieval pipelines from the corpus: chunking, embeddings, retrieval strategy and how answers cite sources.
Scaffolds MCP tools from an OpenAPI contract when an agent needs access to an internal API.
Builds LLM-as-judge and pairwise evaluations so prompt and model changes are compared on evidence.
The ZIP includes a visible installable folder with SKILL.md, skills/, bundled resources and INSTALL.md for your selected agent.
The pack covers AI Application Architecture, Agent Workflow Design, Prompt Evaluation, Retrieval Workflows, AI Integration, Model Output Validation. It combines role instructions, assigned skill files and their bundled resources into an export for your AI agent.
Choose your AI agent on this page and download its ZIP. Extract the package, read its README and source licenses, then follow the install instructions for that client. Tasks may require your own data, integrations or tools.
Yes. The public complete pack can be downloaded without payment or an account. Creating and saving your own role requires a verified account.
No. This is a downloadable configuration for an AI agent, not a hosted employee or a guarantee of correct work. You provide the context and permissions and review the results.