Context Design for a Claude Cowork Sales Agent
Context engineering for a Claude Cowork sales book: what to include, what to exclude, compaction, and sharing a playbook without bloating every sub-agent's context.
Agentic AI, LLM engineering, and the models behind modern automation — multi-agent systems, LLM evaluation and comparisons, RAG, fine-tuning, AI infrastructure, security, and production AI engineering.
From the blog
Context engineering for a Claude Cowork sales book: what to include, what to exclude, compaction, and sharing a playbook without bloating every sub-agent's context.
Context engineering for Claude Code — what to put in the window, what to fetch on demand, how to fight staleness, and why structure beats volume.
Connect MCP servers to Claude Code with scoped auth, strict schemas, structured error handling, and idempotency so tool calls are safe to retry.
Wire MCP servers into a Claude Cowork sales agent: least-privilege auth, typed schemas, structured error handling, and idempotency for a clean 4,000-account book.
Reusable code-level patterns for Claude Code agents — system prompts as contracts, narrow tools, document-tree context, plan-then-act, and compaction.
Reusable patterns for a Claude Cowork sales book: single-job skills, loud idempotent tools, budgeted context, structured handoffs, and evals as regression tests.
Follow a concrete build of a working agent on Claude Code primitives — memory, schema-typed tools, the loop, HTML-framed results, and a real task.
A buildable walkthrough: scaffold skills, wire a read-only CRM connector, run sub-agents, add a write gate, and roll out writes for a 4,000-account book.
How a 4,000-account sales book runs in Claude Cowork: orchestrator, scoped sub-agents, MCP connectors, CRM as state, and a deterministic write gate.